Digital marketing for manufacturers

What digital marketing means for a manufacturer#

Digital marketing for a manufacturer is the work of making sure that when somebody who could issue you a purchase order goes looking for what you make, they find your company and believe what they read. In practice that means a website that states your capabilities as specifications, pages that rank for the processes and materials you actually sell, and a path from a search result to a request for quote short enough that a busy engineer finishes it.

If you have ever sat through a marketing pitch and felt it was written for someone selling shoes, you are right. The phrase comes from consumer marketing, and its assumptions do not hold up on a shop floor. There is no cart and no impulse purchase. A single order can be worth more than a year of consumer sales, and it will be approved by a group of people who have never heard of your company, over a period measured in months.

That is why two words are worth settling early. In most marketing writing, a lead means an email address. On a shop floor, what matters is a request for quote: a real part, a real quantity and a real date, arriving from a company that can buy. One RFQ from a tier one supplier is worth more than several hundred newsletter subscribers, and a program built to grow the second number will quietly starve the first.

Why the standard marketing playbook misfires in manufacturing

Most marketing advice is written for a business where the purchase is small, fast, made by one person and repeated often. On a shop floor, all four of those assumptions are false, and each one breaks a different standard tactic. It is the reason so much agency advice sounds sensible in the meeting and does nothing for your quote count.

What this guide counts as marketing

It helps to draw the line clearly, because a plant with no marketing department is easy to sell things that belong in somebody else's budget. In this guide, marketing means everything that happens before a named person at a named company asks you to price something.

• Marketing covers the website and what it says, being found in search of every kind, the listings and profiles that describe you, case studies and technical content, social media and video, the email you send to people who already know you, and measuring what arrives.
• Marketing ends the moment a quote request arrives. What happens next belongs to sales. We still cover the handoff later, because a program that produces quote requests nobody answers has produced nothing.
• Marketing leaves alone pricing strategy, quoting software, contract terms, and anything that needs a decision about capacity. Those are business decisions that marketing can inform and should not make.

Everything that follows is built around one question: does this make it more likely that a qualified request for quote arrives? Anything that cannot pass that test has been left out, however common it is elsewhere.

Who actually makes the decision#

The person searching is rarely the person who signs. A typical industrial purchase involves an engineer who decides whether a supplier is technically acceptable, a buyer in procurement who runs the quote and owns the commercial terms, and a plant, quality or operations manager who owns the problem and approves the spend. Marketing written for the engineer alone wins the first gate and loses the second.

This is the most common mistake in industrial marketing writing, and we made it ourselves on an earlier version of this site. If you write as if one engineer searches, evaluates and buys, you get content that is technically excellent and commercially useless. It never answers the questions that actually stop a purchase: capacity, terms, insurance, financial stability, and who to call when a line is down.

Read down the third column and you have most of a website specification. Each row is a different page, or a different block on a page. If your site is missing any one of them, it fails that reader, however well it does everywhere else.

How an industrial purchase actually moves#

An industrial purchase moves through six recognizable stages, and marketing only touches the first three. Somebody notices a problem, researches quietly without contacting anyone, builds an informal shortlist, issues requests for quote, validates the winner technically, then settles commercial terms. The whole sequence commonly runs three to twelve months, and by the time your phone rings the shortlist has already been drawn.

That is an uncomfortable thought for any owner. The most important part of the decision happens while nobody is talking to you, and it turns entirely on what a stranger can find and understand on your site without help. You never see the buyers who looked and moved on. If you are invisible during stages two and three, you are only competing for the jobs that happen to reach stage four by accident.

Stages five and six belong to your quality system and your commercial team. What marketing owes them is a place on the shortlist, and a quote request that arrives with enough information to be priced properly.

How long the quiet part actually lasts

The best-documented recent measurement comes from a survey of more than 900 B2B buyers who had made a purchase over $10,000 in the previous two years[7]. It found an average buying cycle of eleven months, of which roughly eight were spent researching before any supplier was contacted, with first contact arriving about 70% of the way through.

Buyers evaluated four suppliers on average, and where six or more were considered, independent research stretched to about 75% of the journey. A survey of engineers specifically puts around 62% of the buying journey online before sales is engaged[8].

The most quoted statistic in B2B marketing, and why not to plan around it

You will meet the claim that 57% of the purchase decision is complete before a buyer contacts a supplier. It also circulates as 65%, 67% and 70%. The figure is usually traced to work by CEB with Google around 2012, and tracing it any further is harder than it should be. The critique that called it a myth could not pin down where it came from either, and its author argues the number is a headline average hiding very wide variation between simple and complex purchases[9].

• The figure has no stable primary source. Four different percentages circulate under the same claim.
• It is an average across very different purchases. A repeat order for a consumable and the approval of a new tier one supplier do not run on the same timeline.
• The share of the decision made and the share of time spent are different things. A buyer can spend most of the elapsed time researching and still have the decision entirely open.
• The measured version is the one to plan with. The 2023 survey above publishes its sample, its threshold and its method, and reaches a similar conclusion on solid ground[7].

This matters for your plan. The 57% claim gets used to justify cutting sales contact entirely, and the measured data says something more useful. Buyers arrive late, and by the time they arrive they have already filtered most suppliers out. The useful response is to be present and easy to understand during the months when nobody is talking to you.

Competing with an offshore price#

Import competition is now a standing condition of quoting. In a 2026 survey of 131 US contract manufacturers, 38% of quotes faced import competition, price was named in 94% of losses, and half of those who lost faced an import price at least 30% below theirs. Marketing cannot close a 30% gap. It can put you in front of the buyers for whom price is one variable among several, and give the ones who want to buy domestically a reason they can defend internally.

Read that survey with its limits in mind. It was published by an organization that exists to promote reshoring, it gives no field dates or sampling frame, and its audience chose to take part[10]. The direction it reports matches what people on quoting desks describe, and the exact percentages are best read as that organization asking its own audience.

The same source reports the other half of the picture: 79% of contract manufacturers had customers raise reshoring, and the share currently quoting reshoring work doubled from 16% to 32%[10]. A separate tally counted 244,940 US manufacturing jobs announced through reshoring and foreign investment in 2024, across 1,412 cases[11]. Those are announcements, and the same organization estimates that a substantial share of the jobs never arrive.

What actually beats a lower price

A buyer who chooses a domestic supplier at a higher unit price is paying for something the quote does not show, and they need help proving it to whoever signs. Publish the things that make that case. They are the same things a specifying engineer wants to see anyway.

Which channels return something, ranked#

For a manufacturer under roughly $50 million in revenue, four channels carry almost all of the return: the website itself, organic search including AI assistants, the Google Business Profile, and email to people who already know you. Paid search is worth it in specific situations. Most of the other channels take more of your attention than they give back, and several are best skipped entirely.

The ranking below is ours, drawn from our work with industrial companies, and the times are estimates. Use it as a starting order. If a channel takes a long time to produce its first quote request, you want a clear reason for it, beyond hope, before it goes into your plan.

The channel most rankings leave out

Referral sits near the top of that table and appears in almost no marketing plan, because it is nobody’s budget line. For a finisher, coater, plater or job shop, the largest single source of work is usually other shops: machine shops and fabricators inside freight range who need an operation they do not run, and competitors declining work they cannot take. It costs nothing, and almost nobody works it on purpose.

• Write down where you send declined work, and check that the shops on that list know they can send theirs to you.
• Publish the operations you subcontract out. A page naming what you send elsewhere tells complementary shops exactly why to call you, and tells buyers exactly where your capability ends.
• List the complementary shops inside your freight radius and visit two of them a quarter. This is sales work with a marketing return, and it is the cheapest hour in the whole guide.
• Add one question to the quote form asking which shop sent the buyer to you. Without it, referrals never show up in any report, and the work behind them is the first thing to lose its budget.
• Price trade press and association directories the same way as any paid listing, by dividing the annual cost by the gross profit on one won job. Engineers rate online trade publications level with supplier websites as a research source[8], so they are worth working out properly before you dismiss them.

Why the ranking looks like this

Three questions separate the top of that table from the bottom. Once you know them, you can rank any channel this guide has not covered.

1. Does it build on itself, or does it reset? A capability page published this year still earns quote requests in three years. An advertisement stops the hour the card declines. Channels that build on themselves are worth starting even when they are slow, because the slow part only happens once.
2. Does it reach people at the moment of need, or before it? Search meets a buyer who already has the problem. Social media meets one who might have it in eight months. Both are useful, and mixing them up is how social media budgets get judged against sales targets and canceled.
3. Is it something only you can do? Anyone can buy the same advertisement. Nobody else can photograph your floor, publish your tolerances or describe the job you saved last March. Channels built on things only you have are the ones competitors cannot copy by outspending you.

Where a channel actually belongs in the sequence

A common and expensive mistake is starting every channel at once, because each one sounds reasonable on its own. They depend on each other, and running them out of order wastes the money spent on the later ones.

• Nothing else starts before the capability content exists. Every other channel points at it. Advertising, social and email all end at a page, and if that page cannot answer a technical question, the money spent getting somebody there is wasted.
• The Google Business Profile can start immediately. It is free, it is fast, and it does not depend on the website being finished. Of everything in this guide it gives the best result for the effort, and it is the one most often left half finished.
• Search work starts as soon as the pages exist, because it takes months to build, and every month of delay pushes the return further out.
• Paid search comes after you know which terms you cannot rank for. Paying for terms you already rank for organically is the most common way to buy traffic you were getting for free.
• Social and video come last, unless you are hiring. In that case they come first, because the people you need to reach are candidates.

The website is the hub, and it has a job description#

A manufacturing website has a specific job: to let a stranger establish, without speaking to anyone, that you can make their part, that you are allowed to make it, and that you have made something like it before. Everything else on the site comes second. A site that does those three things and offers a short path to a quote will bring in more work than a better looking site that does not.

Hold your own site up against those three. Most industrial sites we open handle the third reasonably well with a photo gallery and hint at the second with a certification logo. Most fail the first completely, because the capability information a buyer needs is either missing or locked inside a downloadable brochure that no search engine and no assistant will ever read. If you have hesitated to send a new customer to your own website, this is usually why.

We wanted to know how common this is, so we measured it. Acrosssixty-six US manufacturer home pages, four cited a materials or process standard and thirty-nine named a quality certification. The method, the full dataset and the ten criteria are published in manufacturing website examples, measured, alongside the six sites that do this best and what each of them does differently.

What has to be on your website, as readable text

• Your processes are named the way a buyer names them, and each one has its own page and its own address.
• The materials and grades you routinely work are listed, and the ones you decline are stated as clearly as the ones you accept.
• Your sizes, envelopes and tolerances are given as numbers. This is the most common gap of all, and the easiest to close.
• Your certifications are shown with their scope, including which operating company or site each certificate covers.
• Your capacity and lead time are stated, even as a range, because procurement cannot shortlist a supplier of unknown size.
• The industries you serve are named, because a buyer wants evidence that you have been through their audit regime before.
• A named person, a phone number and a physical address are visible without scrolling.

The quote form deserves its own attention. A buyer should be able to reach it in one click from every page, it should accept a drawing file, and it should ask only for what you need to price the job. Twenty-one of the sixty-six sites we measured fail the first of those three, as the assessment of the best quote form in that sample shows in detail.

Every extra field is another chance for a busy engineer to close the tab. A form that demands a job title and a company size before it will accept a drawing was designed for the seller, and buyers can tell.

The page list a manufacturer actually needs

The structure of the site decides most of how it does in search, and it is decided before a word is written. You want one page for each thing you sell, each at its own address, and each answering one buyer's question completely.

The credentials that decide whether you can be quoted at all

Most industrial marketing treats certification as a logo, but buyers use it as a filter. A buyer with a requirement does not weigh your certification against your tolerances. They remove everyone who lacks it before reading anything else. Name the certifications you hold, with their scope, in the words a buyer searches for.

• Publish the scope alongside the standard. A quality reviewer reads the scope line to see whether your certificate covers the process they need. A certificate number with no scope answers nothing.
• Name the registrar and the expiry date. A certification page showing an expired certificate tells a reviewer that your document control has slipped, whatever the state of your quality system.
• Say where each certification is held. If your group holds certificates for each operating company separately, say so, because a buyer who checks the wrong company will conclude you do not have it.
• If a certification is in progress, say that too, with the target date. Buyers plan around timelines and will sometimes wait.

Write the capability table before you write the page

A good capability page is a table with writing around it, and the table comes first. Draft it with whoever quotes the work, and the rest of the page follows easily. Anything you cannot fill in is either a capability you do not have or a number nobody has written down, and it is far better for you to find that out before a customer does.

• State the range you hold every day. The tightest tolerance you have ever held once is a sales claim. The tolerance you hold across a full shift is a specification, and it is the one that survives a first article.
• State what you decline. A shop that does not run stainless should say so. It costs a few unqualified inquiries and earns credibility with everyone else.
• Name the standard instead of reaching for an adjective. A buyer can evaluate a finish described against a named specification. Nobody can evaluate a finish described as excellent.
• Give the units. Say whether the numbers are in inches or millimeters, because nobody can quote from a table of bare numbers.

Work you cannot publish

Some manufacturers have a genuine reason not to put their work on a website: customer confidentiality agreements, defense work under export control, or a part that identifies a customer's unreleased product. The answer is to publish the capability without the customer, which is almost always possible and almost always skipped.

• Describe the process, the material, the tolerance and the volume, and leave the customer and the part unnamed.
• Say plainly that you work under confidentiality and will discuss specifics under an agreement. To a buyer with sensitive work, that sentence is a credential in itself.
• Where the work is export controlled, say which regimes you operate under and what registration you hold. Buyers looking for that capability search for exactly those words.
• Get written permission before you name any customer. A case study published without it can cost you the account it describes.

Publishing the technical data engineers need#

Engineers look for downloadable technical data before they ask for a quote, and the supplier who publishes it gets evaluated while the one who asks them to call does not. Two surveys of engineers put the effect of published CAD models on supplier selection very high. Both come from companies that sell CAD publishing services, so keep that in mind as you read the numbers.

A survey of 2,563 designers and engineers found 92% saying the availability of a CAD model influences their choice of supplier[12], and a separate report claimed 82% of downloaded CAD models led to a purchase[13]. Both figures come from CAD content platforms, and the second gives only a count of companies, with no count of respondents at all. Read them as an industry making its own case. The case is still probably right in direction, because it matches what engineers say when less interested parties ask them[14].

What to publish, in the order it gets used

1. Publish specifications as web pages first, and as PDFs second. A web page can be found, indexed and quoted. The PDF is what gets attached to an internal email. If you publish only the PDF, you get the second benefit without the first.
2. Publish your material and process data. That means grades, tempers, hardness after treatment, thickness ranges, and the standards each one is measured against.
3. Publish CAD models if you sell a part. A job shop making to print has nothing to publish here, and a component maker has everything to publish here.
4. Say which file formats you accept, and which you prefer to receive. This one line saves a whole round of email on every first inquiry.
5. Publish your drawing and inspection templates. An openly published first article report template or print template is quietly one of the most linked-to things a supplier can offer.

The quote path, and what happens after the request arrives#

The quote form returns more for the work than anything else on a manufacturing website, and it gets the least attention. Two decisions shape most of what it brings in: how few fields it asks for before it will accept a drawing, and how fast a person replies afterward. A program that produces quote requests nobody answers quickly has produced nothing, and the evidence on response speed, though old and thin, is still hard to argue with.

You may already suspect that requests sit too long in someone's inbox. The most quoted study on this sent one web inquiry to each of 2,241 US companies and measured the reply. The average first response took 42 hours, 23% never replied at all, and firms responding within an hour were about seven times more likely to have a meaningful conversation with a decision maker than those responding in the second hour, and sixty times more likely than those taking a day or more[15].

That study is the basis of what slow and unanswered quote requests cost over a year, which puts your own volume and win rate against the same benchmark. The calculator leaves out a decay curve, because the response-time multipliers that circulate online come from a different context.

The study itself is a two-page magazine item with no methods section, no field dates and no industry breakdown, and it is fifteen years old[15]. We use it because the direction is almost certainly right and it is still the best evidence there is. The question deserves a better study than it has had.

The form: ask for less than you want

A benchmark of more than 250 large sites found an average of 11.3 form fields where about 8 would do[16]. That study looked at consumer checkouts, and the lesson carries over to industrial quoting: every field is a point where a busy person can stop.

• Accept a file first. The drawing is the request. A form that will not take an attachment forces a second email, and you lose everyone who never sends it.
• Four fields are usually enough. Ask for the file, the quantity, the target date, and how to reach them, and ask everything else in your reply.
• Never ask for company size or job title before the drawing. Those fields help your reporting and do nothing for the buyer's request.
• Put a named person and a direct number beside the form. Some buyers will always prefer to call, and the ones with a stopped line always will.
• Say what happens next and how fast. A stated response time is a commitment, and it is also the most persuasive sentence on the page.

After the request arrives

1. Acknowledge it inside the hour, even if only to say it arrived. An automatic reply naming the person who now owns the request costs nothing and gives you the time to quote properly.
2. Send it to a shared inbox that more than one person reads. Someone's vacation should never be the reason a quote goes unpriced.
3. Log it the same day, along with where it came from. Without that log, none of the measurement in this guide is possible.
4. Reply to the ones you decline. A clear no, with the reason, is remembered, and it often comes back later as the right job.

Search now happens in two places at once#

Search for industrial suppliers now happens in two places that read the same pages. Google still sends the clicks and rewards pages matching a specific query. Assistants such as ChatGPT, Perplexity and Google AI Overviews read those same pages and return a written shortlist, often naming three or four suppliers before the buyer sees a results page at all. Work that earns one tends to earn the other.

The practical difference is what a win looks like. On Google, a win is a position and a click. Inside an assistant, a win is being named, and the buyer may arrive at your site already believing you can do the work, or may never visit at all and simply put you on the list. That second kind of win never shows up in your analytics, which is one more reason to count quote requests instead of visits.

Google makes the same point itself. Its generative features, including AI Mode, are built on the same core ranking systems and draw from the same index, which is why the company's own summary of the era is that good SEO is good GEO. [19] That is Google describing its own products, and it is still the clearest signal available that the two kinds of work are really one.

What the measurements show, and why the two audiences differ

Pew Research tracked the browsing of 900 US adults through March 2025 across 68,879 Google searches, 12,593 of which produced an AI summary[20]. Where a summary appeared, people clicked a traditional result 8% of the time, against 15% where none appeared. Only 1% clicked a link inside the summary itself, and 26% ended their browsing session there against 16% without one.

Read alone, that suggests people are clicking through less and less. Set against the surveys of engineers, it says something more precise. Among technical buyers, 75% have noticed AI summaries, 69% read the summary and then continue into the actual results, and only 6% say the summary is usually enough information[14]. Trust in AI answers among engineers averages 4.7 out of 10, and only 10% rate AI tools as very or extremely trustworthy content authors, against 66% for engineering experts[14].

One more figure from the Pew study is one you can act on straight away. Summaries appeared on 18% of searches overall, but on 60% of searches that began with a question word, and on 53% of searches of ten words or more[20]. Long searches phrased as questions are exactly what a buying group types, which is the practical case for giving each buyer question its own page and its own heading. That is why this site publishes twenty of them as separate pages instead of one list.

Specific beats broad, and the arithmetic proves it

If you have ever searched for the process you run every day and found competitors, directories and companies two states away, this is where that gets fixed. A category term such as machine shop has large volume and attracts students, competitors and people three states away. A term naming a process, a material and a state has a fraction of that volume, and almost everyone who arrives on it has a part to make. Ranking second for the narrow term brings more quote requests than ranking eleventh for the broad one, and the narrow term is within reach this year.

The habit underneath all of this is old enough to be well documented. Google research with about 3,000 B2B researchers found 71% beginning with a generic search instead of a branded one. In other words, they search for the thing they need before they search for anyone who makes it[21]. That study was fielded in 2014 and has no separate figures for manufacturing, so trust the direction and treat the number as dated[21].

How to choose the terms, without buying a tool

Keyword research software is useful, but it is the wrong place to start. The terms that bring a manufacturer quote requests are usually too specific to show up in a tool with any volume attached, so people throw away exactly the right words because of the low number next to them.

1. Read your last twenty quote requests. Copy the words the customer used to describe what they wanted. These are your terms, written by the people who buy.
2. Ask your estimator what people say on the phone. The way people describe a job on a call is closer to how they search than anything in your brochure.
3. Combine three things in each term. Use a process, a material or industry, and a qualifier such as a region, a tolerance class or a volume. Terms with all three bring in requests, and terms with only one rarely do.
4. Search each one and look at who ranks. If the results are directories and competitors, the term is commercial. If they are encyclopedia entries and university pages, it is an academic term and it will bring you students.
5. Drop anything you would not want to quote. Ranking for work you do not want is worse than not ranking at all, because it eats estimator time you cannot bill.

This is why the way your pages are organized matters more than how many words they have. One page listing eight capabilities ranks for none of them. Eight pages, each naming one capability with its own tolerances and its own examples, carry the same information arranged so buyers can find it. How we do the search work follows exactly that pattern.

How a manufacturer gets cited by an AI assistant#

An assistant names a supplier when three conditions hold: the claim is stated as plain text on a page it can read, the company is consistently identifiable across the web, and something outside your own website corroborates the claim. Most manufacturers fail the first condition for capability data and the third condition entirely, which is why assistants keep naming the same handful of larger firms.

None of this needs new technology. It needs the facts a buyer would ask for to sit somewhere a machine can read them, written so they still make sense when quoted on their own.

What actually helps

1. State facts in text on the page. A capability matrix saved as a JPEG or locked in a PDF is invisible to an assistant. The same matrix as an HTML table can be quoted.
2. Give questions their own addresses. A page whose heading is the exact question a buyer asks is the shape an assistant quotes. We publish twenty of them, one question per page.
3. Publish the numbers you are willing to stand behind. That means prices, lead times, tolerances and dates. Vague pages get summarized, and specific pages get quoted.
4. Keep your name, address and phone number identical everywhere. Assistants work out which company is which from those details. Three spellings of your company name look like three weak companies instead of one strong one.
5. Get mentioned away from your own site. Directory listings, review platforms, association memberships, trade press and customer pages all count. What you say about yourself carries little weight on its own.
6. Let the assistants read your site. Check that your robots file, the small file that tells crawlers what they may read, does not block GPTBot, PerplexityBot, ClaudeBot or Google-Extended. Many sites block them by accident through a default setting.

What being cited is measurably worth

Two independent analyses put numbers on both halves of this. An analysis of 300,000 keywords found click-through at the first organic position running about 58% below forecast where an AI Overview appeared[22]. That is a correlation, measured against a projection instead of a controlled comparison[22]. A separate study of 3,119 terms across 42 organizations found that brands cited inside the AI answer saw organic click-through of 0.70% against 0.52% when they were not cited[23], on a convenience sample of one agency’s clients[23].

Read together, they describe the same market. The results page gives up clicks, and being named inside the answer wins back a meaningful share of them. Neither study settles it alone, and both point the same way.

How many slots there are, and who fills them

Pew found that 88% of AI summaries cited three or more sources[20], and that the median summary ran 67 words. Read the first figure carefully. Three or more is a floor, and Pew publishes no typical count, so think of three as the fewest suppliers you are competing against, and leave open how many slots there really are[20].

The second figure is the one to act on. Sixty-seven words is roughly one specification, one qualifier and one number, so your claim has to live in a sentence that still works when it is lifted whole off your page.

Being named also builds on itself, and that has been measured separately. Among engineers, 70% say that where two solutions are technically similar they are likely to choose the better known brand, and 53% say brand familiarity influenced their most recent purchase[14]. Only 34% report buying purely on specification, price or performance. Technical parity is more common than most manufacturers admit, and recognition breaks the tie.

You can test this yourself. Ask each assistant the questions your buyers ask, in their words, and write down which companies get named. That list is your real competition for the next few years, and it is often different from the one you would have guessed.

How to run that test properly, in an afternoon

Assistants have no equivalent of Search Console, the free Google report that shows how your site appears in search. Nobody outside OpenAI and Google can see what people typed, so the only way to measure it is to ask the questions yourself and record the answers. Done once, it is an anecdote. Done the same way each quarter, it becomes a trend line, and it costs you an afternoon.

1. Write twenty prompts in your buyer's words. Five shapes cover most of it: the capability with a material and a tolerance, the capability with a location, a certification plus an industry, a near-competitor plus the word alternatives, and the plain question a buyer asks a colleague. For a plating shop that is closer to anodizing supplier for aerospace brackets in Ohio than to metal finishing.
2. Ask in a logged-out private window. Otherwise the assistant will show you a flattering answer built from your own history. Repeat each prompt two or three times, because the answers vary between runs and one reading tells you very little.
3. Ask for the sources explicitly. If you add a request to list every source URL used, the citations come out where you can record them. The sources matter more than the answer, because they are the pages you can realistically get onto.
4. Record two columns: who was named, and what was cited. A spreadsheet is enough. The companies named are your real competition. The sources cited are your target list, and most of them will be directories, trade publications and forum threads, with few supplier websites among them.
5. Find the gaps and work on them. Sources that name a competitor and never you are the shortest way in. Some will add a listing or make a correction if you ask. Where you cannot get onto a source, you can publish something that follows the same pattern, only better, on your own site, which is the subject of the section on earning links.

One column of that spreadsheet has just become easier to fill. Search Console has begun reporting impressions from Google's AI features, so you can now partly see your AI Overview appearances in a report as well as by asking. It only counts Google's own features and leaves out the other assistants, so the manual check is still worth the afternoon.

Being cited and being recommended are different outcomes#

An assistant citing your page as a source and an assistant recommending your company are separate events, and the gap between them is wide. In one hundred business-to-business best-category queries measured on Google AI Overviews, a company that had written the cited list article was left out of the recommendation itself about 69% of the time. The practical consequence is that a best-in-category article naming your competitors can be read as a menu, and the reader orders from it. Whether that matters depends entirely on whether an assistant already names you.

You may have been pitched this tactic already. It goes like this: assistants favor list articles, so publish one for your own category, put yourself at the top, and name your competitors underneath to look even-handed. The first half of that is true. The second half is where it goes wrong, and it has now been measured twice.

One hundred best-category queries were tracked against Google AI Overviews and AI Mode at three checkpoints in spring 2026. Where a company’s own list article was cited as a source for the answer, that company was missing from the recommendation in 224 of 323 cases, close to seven times in ten[28]. The page did its job as research material and returned nothing to the company that paid for it.

The condition that decides which way it goes

A second experiment, smaller and run by a vendor on its own two brands, found the part the first one cannot see. It tracked thirty-four self-promotional pages against 9,886 answers across four assistants. For a brand that assistants were not naming at all, 82% of its new appearances came from its own cited pages. For an established product in the same account, the same tactic accounted for 6%[29].

Those two results fit together. If nothing names you today, your own page is a way into the answer, and a citation that stops short of a recommendation is still a first appearance. If you are already named, a page listing eight competitors gives an assistant eight alternatives to your own entry, sourced from you. The question to settle first is which of those describes your company, and it takes about ten minutes.

If you publish one anyway

Sometimes the page is worth having for ordinary search reasons, and the assistant question comes second. The advice below follows from the results above. Nobody has published a controlled test of it yet, so treat it as a sensible reading of those results.

• Name three competitors at most. Every name on the page is another supplier an assistant can pick up, described in your own words. Search rewards being thorough, and here that same instinct works against you.
• Put yourself first if you are going to be on it. If you list a competitor at the top to look even-handed, you are telling a model, in plain text, which supplier you consider best.
• Write it as an honest editorial piece. A page with a request-a-quote button after every entry reads as a sales page that happens to mention competitors, and it gets treated as one.
• Be careful listing your own shortcomings beside a rival. A weakness stated next to a competitor’s name tells the assistant who to recommend whenever that weakness matters.

What to publish instead, in either case

Both studies point the same way on this, and so does everything earlier in this guide. The formats that work are the ones where you are the subject of the page, instead of one item on a list.

• Publish a page for each thing you actually make. When an assistant answers from a page about your process, material and tolerance, it can only do so by naming you, because the facts on it are yours.
• Publish a head-to-head comparison where a real decision exists. Weigh the two options properly, including the cases where the other one wins. Our own are a marketing hire set against outside help and the four supplier types compared on seven axes. Neither names a firm, because the choice a buyer is making is between kinds of supplier.
• Give each buyer question its own page. Answer it at its own address, with the answer first. That is what the question pages on this site are, and it is the cheapest structure to build.
• Publish your own measurements. An assistant cannot find a number nobody else has on a competitor's page. Our study ofsixty-six manufacturer websites exists for that reason, and the dataset is published with it.

Keep one caveat in mind. Both studies measured software categories, one of them on a single platform, and the smaller one was a vendor measuring itself on thirty-four pages. Neither measured industrial suppliers, the sources assistants draw on shift from week to week, and nobody outside Google can say what separates a citation from a recommendation. What holds up is narrow and still useful: being read is a different thing from being chosen, and a page that lists your competitors describes them to a machine.

Google Business Profile, the most under-used asset a plant owns#

A Google Business Profile is free, takes an afternoon to complete properly, and is the asset manufacturers most often leave half finished. It is also where the standard advice is most clearly wrong. The evidence says posting often does very little for ranking, while the categories you choose matter more than anything else, and most plants put their effort into the first and never revisit the second.

Start with what Google itself says. It states that local results are mainly based on relevance, distance and popularity, and its own list of ways to improve local ranking covers verification, complete and accurate business information, categories, replying to reviews, and photographs. Posts are not mentioned anywhere on that page[30].

Instead of judging your profile by eye, score it against the fields that make a difference. Thirteen checks over a hundred points weights the primary category far above everything else, for the reason given above, and takes about ten minutes with your profile open beside it.

The strongest independent evidence points the same way. In a survey where 47 local search practitioners scored 187 named ranking factors, the frequency of Google Posts ranked 148th of 187 for local pack ranking and 186th of 187 for local organic ranking, and appears on the report’s own list of local search myths[31]. The primary business category ranked first of 187.

Reviews, and the part most manufacturers get backwards

Most manufacturers think of reviews as something for restaurants and feel awkward asking a customer for one. Two things suggest it is worth getting past that. Google names reviews and replies in its own list of things that help local ranking[30], and a survey of 1,002 consumers found 74% saying they only care about reviews written in the last three months[32].

That second figure comes from a consumer panel, so the principle carries over to industrial buyers while the exact number may not[32]. The principle is the useful part: forty reviews from 2019 are worth less than four from this quarter, which turns reviews from a one-time campaign into a small habit you keep up.

The profile checklist, in the order that matters

1. Choose the primary category first. Pick the most specific one that describes what you sell, then search your own capability terms and see who appears. If competitors with a narrower category outrank you, the category is the reason.
2. Add secondary categories, one for each genuine service line. Choosing a few accurate ones works better than guessing wide.
3. Make the name, address, phone number and hours identical everywhere. Use the exact legal name, one number that a person answers, and hours that match your shifts.
4. Add photographs of the actual plant. They rank 96th, and they are still worth doing, because a buyer deciding between two listings is checking whether you are real.
5. Ask for a review after every job that went well, and reply to all of them. A small, steady flow works better than a burst.
6. Then post, knowing that posts are for the person who has already found you, and do little for your ranking.

We have done this many times, and it is the cheapest work in this whole guide and almost never finished. It gets abandoned because posting, the part everyone can see, does the least. The effort seems to produce nothing, and the profile gets written off. Doing the first four items once is worth more than a year of posts.

Industrial directories and platforms, and where they fit#

Industrial directories and quoting platforms can be worth a listing, but they are no substitute for a strategy. They put you in front of buyers you would not otherwise reach, and they do it by renting you visibility for your own capability, on a page you do not control, next to your competitors. Decide with arithmetic: what the listing costs against what one won job is worth.

The scale is real, and it is smaller than their marketing suggests. One quoting marketplace disclosed in its annual SEC filing that it had 81,821 active buyers and 4,996 active suppliers at the end of 2025, with 1,760 accounts spending at least $50,000 in the trailing twelve months[33]. That is roughly sixteen buyers per supplier, and a small minority of accounts carrying serious volume.

How to decide about any one of them

1. Ask your customers where they looked. One question on the phone is enough. If nobody names the platform, the platform does not have your buyers.
2. Do the arithmetic before the sales call. Divide the annual listing cost by the gross profit on one won job, and you have the number of jobs it must bring in. If that number is above two or three, be skeptical.
3. Check what it does to your margin. A platform that sits between you and the customer and runs a reverse auction is selling you volume at a price you do not set.
4. Check whether it outranks you for your own capability. If a directory page for your process outranks your page for your process, you are paying to appear on a page that took your position.

The bigger point is ownership. A directory listing is rented, and it stops the day you stop paying, while a capability page on your own domain is paid for once and stays yours. Use directories for reach you cannot earn yet, and let every month you pay for one remind you to publish the page that will one day make it optional.

Getting onto approved vendor lists#

For tier one, aerospace, medical and defense buyers, there is a gate that search visibility does nothing to open. Before you can be quoted, you have to be an approved supplier, and approval is a paperwork exercise run by quality and procurement. Being easy to find gets you asked for the supplier packet. Having the packet ready is what gets you on the list.

This is where a great many capable shops stall. The buyer is interested, the engineer is satisfied, and then a supplier questionnaire arrives asking for eleven documents, three of which nobody can find. Six weeks pass, and the job goes to the shop that answered in two days.

The packet, assembled once and kept current

• Keep a one-page capability statement. It lists your processes, materials, sizes, tolerances, certifications with scope, capacity, locations, and the contact for quotes, on one dated page as a PDF.
• Keep your certificates with their expiry dates, along with your quality registrations and the registrar for each.
• Keep certificates of insurance showing the coverage levels your customers typically require, which may be higher than the minimum you carry.
• Have a W-9, a D-U-N-S number and your NAICS codes ready. They are all easy to get, and each one can cost you two weeks when someone asks for it out of the blue.
• Keep excerpts from your quality manual, usually the scope, the control of nonconforming product, and calibration.
• Name one person who owns the packet, and give them a calendar reminder to review it twice a year.

Where to register

• Register on prime and OEM supplier portals. Most large buyers run their own, and registering is often the only way to be invited to quote at all.
• Register on SAM.gov and get a CAGE code if you want federal work, directly or as a subcontractor. It is free and slow, and it gets you into the running without giving you an edge.
• Register for any socioeconomic certification you actually qualify for. Small business, veteran owned, woman owned and HUBZone are set-aside categories that buyers filter on, and when buyers filter, you are either on the list or invisible.

Earning the links and citations that carry everything else#

A link from another site is still how a search engine decides which of two equally good pages to trust, and it is the part of the work an agency cannot do for you cheaply. For a manufacturer the sources are unusually specific and mostly unclaimed: the trade associations you already pay dues to, the distributors who already sell your product, the customers who publish approved supplier lists, the standards bodies whose scopes you already hold, and the local and trade press. Publishing something worth citing works better than asking, and the assets that earn citations are the ones only a plant can produce.

This is the least glamorous part of the subject and the one most often skipped, usually because the advice out there is written for a coffee blog or a software product. Links work the same way for a plant. What differs is who is likely to link to you, and a manufacturer starts that list further ahead than almost any other kind of business, because most of the relationships already exist and nobody has ever asked them.

Coverage first, outreach second

The order matters, and there is now measurement behind it. A study of 173,902 URLs found that a page ranking for the main query and at least one of the sub-queries an assistant splits it into was 161% more likely to be cited in an AI Overview than a page ranking for the main query alone[35]. The same study found that roughly 68% of pages cited in AI Overviews did not rank in the top ten for any of those queries. That is the more encouraging half for a small supplier, because covering a narrow subject in depth can earn you a citation that ranking never did.

Keep its limits in mind. It is a correlation, and the authors say so. The sub-queries also change so much that only about a quarter come up again when the same term is searched twice. What holds up is a principle to work by, and you should not treat it as a target list. Covering one subject from every angle beats covering ten subjects once each, and it also gives an outreach email something to point to, because there is a page on the other end of the link worth a reader's time.

Who actually links to a manufacturer

Almost none of these need cold outreach. They are relationships you already have, and nobody has asked them for a link yet.

The three things worth publishing to be cited

• Publish measurements only you can take. That might be a tolerance capability study across a material range, cycle time against wall thickness, a failure analysis with photographs, or scrap rate before and after a fixture change. A number that exists nowhere else cannot be found on a competitor's site, and it is the one kind of content a language model cannot make up. It is also the reason our own study ofsixty-six manufacturer sites was run and its dataset published.
• Publish tools, converters and templates. The calculation your estimator does forty times a week is a page other people will link to for years, and anyone technical can now build one in an afternoon. Every calculator on this site exists for that reason, and each one publishes its formula so anyone can check the arithmetic.
• Publish the document your customers keep asking for. It might be a material selection guide, a drawing checklist or a finish comparison. If your estimators email the same explanation again and again, that explanation should be a page, and the people asking for it are the people who will link to it.

There are two things to turn down. Paid link placements and bulk directory submissions are still sold to manufacturers by the thousand, and they are the one part of this subject that can actively hurt you as well as waste money. Guest posts written for the link, with no thought for the reader, are also now easy to spot. The approach that works is the dull one: build something worth citing, then tell the eight groups above that it exists.

The content engineers and buyers actually use#

The content that brings in quote requests is permanent, specific and a little boring: one page per process, one per material, one per industry, case studies with numbers in them, and direct answers to the questions buyers ask before they call. Company news, anniversary posts and general industry commentary produce almost nothing, and they use up the time the useful pages needed.

Before you write anything, ask two questions. Will this still be true in three years, and would a buyer searching at two in the morning be glad it exists? Pages that pass get a permanent address. Pages that fail belong in an email or a LinkedIn post, where their short life does no harm.

How to write a case study a buyer believes

Name the part or the problem, the constraint that made it difficult, the process chosen, the tolerance or specification held, and the measured outcome with a number attached. Get written permission for anything the customer could consider confidential. If they say no, keep the case study and describe the work without naming the customer.

What the survey data says about content that works

In the Content Marketing Institute's manufacturing study, 104 manufacturing marketers rated video their most effective format at 74%, followed by case studies at 45%, while the most heavily used format was the short article at 89%[36]. The real finding is the gap between what gets made and what works. Only 20% rated their content strategy very effective, and the reasons given were that it was not tied to the customer journey (47%), not data driven (46%) and lacked clear goals (40%).

Before you book a booth, price the one you ran last year against the work it actually won. A booth costed against the jobs it produced counts the stand, the freight, the travel and the staff days, and on its default figures it shows a loss. Put your own numbers in and argue with the result before the deposit goes in.

On gating, where our own position had to move

We previously advised never gating technical content. The engineer survey complicates that: 85% say they are willing to provide their details for technical content they find valuable[14]. Asking for details does not drive this audience away the way it is often said to.

The cost shows up somewhere else, and it is still real. A page behind a form cannot be indexed, cannot be quoted in an AI summary, and cannot appear in the three or so citation slots described above. So our advice now is narrower than before. Publish the page openly so it can be found and cited, and offer the deeper material, such as the full drawing pack or the complete capability document, behind a form for the people who want it. You lose nothing, and buyers can still find you.

Where the material already exists

If you are thinking nobody has time to write all this, the good news is that most of it is already written. Almost every manufacturer already has most of the content this guide recommends, in places nobody thinks of as marketing, so the work is mostly copying it out.

• Start with the quotes you have sent. The capability statements in them are already written, already accurate, and already in your customers' language.
• Collect the questions your estimator answers on the phone. Each one is a page, and the answer already exists in spoken form.
• Look at the inquiries you turned down. The reasons you turn work away mark exactly where your capability ends, which is what a buyer needs to decide whether you are a fit.
• Open your quality manual. Certification scopes, inspection methods and equipment lists are already documented there, for an auditor who reads more harshly than any customer.
• Reuse your training material. It was written to explain the process to somebody who does not know it, which is the same job a capability page does.

Social media for manufacturers, ranked#

For a manufacturer the order is Google Business Profile first, LinkedIn second, YouTube third, and everything after that optional. The profile is the only one that reliably brings in inquiries on its own. LinkedIn mostly reaches procurement, plant management and future hires, with fewer engineers, and that is a useful audience once you plan for it. The rest are worth it mainly for hiring.

It is easy to expect social media to bring in industrial work directly, and it rarely does. Its realistic jobs are to back up what your website says, to help you hire, and to keep you in the mind of somebody who will have a problem in eight months. Judged against those jobs, it is worth doing. Judged as a source of quote requests, it will disappoint.

1. Your Google Business Profile comes first. It is free and fast, and it is the only entry on this list that regularly brings in calls by itself. Complete the category, hours, service area, photographs and description, and keep the details identical to the website.
2. LinkedIn comes second, because that is where plant managers, quality managers, procurement and future hires spend time. Post the work: a fixture you built, a part you held to a hard tolerance, a machine arriving. Company news does poorly there for the same reason it does everywhere.
3. YouTube comes third. A process video answers capability questions faster than any page of text, and the videos keep being found for years. A phone on a tripod is enough.
4. Instagram and Facebook help most with hiring, and with your standing in the local community after that. Both matter more than marketing writing usually admits when you are competing for machinists.
5. X matters little for this audience. Do it if somebody on your team enjoys it, and leave it if nobody does.

The survey data supports this ordering. Among manufacturing marketers, LinkedIn is rated the most valuable social platform by 85%, with YouTube at 40% and Facebook at 31%[36]. Engineers themselves name YouTube and LinkedIn as their most valued platforms[14]. Nothing else comes close for this audience on either side of the transaction.

What to post when you have nothing to announce

Manufacturer social accounts go quiet because the plant is waiting for news, and news is rare. The material is already there on the floor, and none of it needs anything special to have happened.

• Show a part, before and after. It is the most reliable post in industrial social media, and it takes two photographs.
• Show a fixture somebody built to solve a problem. Other engineers find this genuinely interesting, which is a low bar that most corporate content still misses.
• Show a tolerance held, with the gauge in the shot. It is proof that needs no explanation.
• Film a machine running for fifteen seconds, with no music. The sound and the motion do the work.
• Introduce a new hire in their first week, with their permission. This is recruiting content, and it reaches the same people who might buy from you.
• Share something that went wrong and what you changed. It is rare and uncomfortable, and it is the most credible thing a supplier can publish.

Three posts a week from the floor do more than a polished monthly campaign, and one person with a phone can keep that up. That cadence is what our own social media service is built around, at $1,900 a month.

Video, which this audience rewards more than any other format#

Video is the format this audience rewards most and the one manufacturers produce least, and that gap is your opportunity. Manufacturing marketers rate video their most effective content type, engineers name YouTube among their most valued platforms, and almost none of it needs to be expensive. A phone on a tripod pointed at a process is more persuasive than a produced brand film, because it shows the work happening.

In the manufacturing content survey, video was rated most effective by 74% of respondents, while short articles were the most produced format at 89%[36]. That is a sample of 104 marketers, half of them at companies with a thousand or more employees, so read it as a signal and leave room for exceptions[36]. The direction agrees with what engineers report about where they spend their attention[14].

Six videos worth making, in order

1. Film a process running, uncut. Keep it to thirty to ninety seconds, with real sound and no music. This is the video that answers the question of whether you can actually do this.
2. Film a walk through the facility. Two minutes on a phone, narrated by whoever runs the floor, stands in for the site visit a buyer cannot make.
3. Film a measurement being taken. Someone putting a gauge on a finished part is the most convincing quality claim you can make.
4. Explain one design-for-manufacture point. Pick one common mistake on the prints you receive and show how to avoid it. This earns links and goodwill from the engineers who send you work.
5. Film the plant for recruiting. Cover the shift patterns, what the work involves, and who a new hire would report to. This has the highest return of anything on the list, and it serves two departments at once.
6. Film a customer explaining a problem you solved, once you have their permission. It is hard to get, and it is worth more than the other five combined.

Email and automation, sized for a plant with no marketing department#

Email works for manufacturers in one specific form: occasional, useful messages to people who already know you, sent by a named person. A quarterly note to past customers and quoted-but-lost prospects, saying what you have added and what you are seeing, will do better than any automated sequence a plant of this size can realistically keep running.

Marketing automation platforms are built for companies with a marketing department. If one person handles your marketing alongside three other jobs, a spreadsheet and a calendar reminder will do more for you than a platform whose subscription costs more than the campaigns running inside it.

• Write to people who have met you. That means past customers, quoted-and-lost prospects, people you spoke to at trade shows, and applicants.
• Send four times a year, from a named person. Write about a new capability, a new certification, capacity that has opened up, or something you learned on a difficult job.
• Keep it short enough to read on a phone between two meetings, because that is where it will be read.
• Never buy a list. Sending cold email in volume from your main domain puts at risk the quotes and purchase orders that domain carries, which is a large downside for a small upside.

The four emails worth writing

You do not need a content calendar. You need four messages a year that a customer is glad to receive, and each one has an obvious trigger inside the business.

Stop reporting open rates

Apple clients accounted for 62.26% of measured email opens in a single month of more than a billion[38]. That bucket includes Mail Privacy Protection, which fetches images automatically whether or not a human looked[38]. An open rate now mostly tells you which mail client your list uses. It says almost nothing about interest, and in a report it actively misleads.

Report replies and clicks through to a capability page instead. Both need a person to actually do something, and on a list of a few hundred industrial contacts the numbers are small enough to read one by one, which tells you more than any rate.

Deliverability, in the only detail that matters

Your email carries quotes, purchase orders and drawings. Protecting that is worth more than any campaign, and three settings on your domain do most of the protecting. Ask whoever runs your IT whether all three are set up, because the answer is often no.

• SPF lists which servers are allowed to send email as your domain.
• DKIM signs your mail so the receiver can prove it was not altered.
• DMARC tells receivers what to do when the first two checks fail, and reports who is trying to send as you.
• Send nothing in bulk from the domain that carries your quotes. If a marketing platform is sending as your main domain, a spam complaint rate that would be unremarkable for a newsletter can start affecting whether your purchase orders arrive.

Paid search, and when it earns its place#

Paid search is worth it in four situations: defending your own company name, filling a capacity gap this quarter, launching a capability you have no search history for, and reaching a region you cannot rank in yet. Outside those four, money spent on ads before the website and the capability pages are right mostly pays for visitors who leave.

The way it goes wrong is well known and still common. A broad match campaign on a category term spends a month of budget on students, competitors and people looking for something else, brings in a handful of inquiries nobody can quote, and confirms the owner's suspicion that none of this works. If that has happened to you, the problem was the setup, and the rules below fix most of it.

Rules that make the difference

• Use exact and phrase match only until the search terms report shows you can afford anything looser.
• Write ad copy that puts off the wrong clicks. Naming a minimum order quantity or a materials list in the ad text costs you impressions and saves you money.
• Add negative keywords from day one, such as jobs, salary, courses, free, DIY, wholesale, and the names of products you do not make.
• Send clicks to the matching capability page, with the full navigation in place, so a serious visitor can go deeper.
• Spend at least about $1,000 a month. Below that there is rarely enough data to improve anything, so the money buys clicks without teaching you what works.

Defending your brand name deserves its own mention. If a competitor bids on your company name, a small campaign that always runs on your own name is cheap insurance, because the person searching has already decided to find you.

What a competent campaign looks like on day one

Most of the waste happens in the first two weeks, before anyone has enough data to notice. A campaign set up this way loses far less of it.

1. Run one campaign for each process. Budgets are set per campaign, and a single campaign for the whole company spends its money wherever clicks are cheapest, which is rarely where they are worth most.
2. Set the geography to where you can actually ship profitably. A heavy part has a service radius, and paying for clicks outside it is paying to be told no.
3. Match the ad schedule to when quotes get raised. For most industrial work that is business hours in the buyer's time zone, and overnight spend is close to pure waste.
4. Set up conversion tracking on the quote form before the first click. A campaign that runs for a month without tracking cannot be judged, and that month cannot be recovered.
5. Read the search terms report every week for the first month. This is where you find out what Google thinks you sell, and it is often wrong in expensive ways.

Trade shows, and the stage everybody skips#

Trade shows are expensive, slow to repay, and rated by manufacturing marketers as their single most effective channel, and all three are true at once. The research is also unusually clear about where things go wrong. Academics have studied the run-up and the booth heavily and the follow-up barely at all, and the follow-up is exactly where most exhibitors lose the money.

Manufacturing marketers rate in-person events their most effective distribution channel at 51%, ahead of blogs and webinars at 49%[36], and 75% of engineers said they planned to attend at least one in-person event[8]. Against that, US business-to-business exhibitions in one industry index were running 8.4% below the equivalent quarter of 2019, with exhibiting company counts down 8.8%[40].

The academic model that has held up longest treats show performance as three separate stages: attracting the right attendees to the stand, making contact, and converting that contact afterward[41]. Firms tend to budget almost entirely for the middle stage, which is the booth.

What to do instead, at the same cost

• Book the follow-up time before you book the booth. Put two protected days in the calendar for the week after, for the person who will make the calls.
• Write the follow-up email before you travel, while you still have the energy to make it specific.
• Publish something the week before the show naming the show and what you will be demonstrating. It takes an hour, and people planning their route around the floor will find it.
• Photograph and film at the show. You can make a year of social content in two days that would otherwise never be shot.
• Log every conversation the same evening, with the company and what they wanted. A badge scan with no note attached is worth close to nothing eight weeks later.

In short, a show is worth doing when you have the time to follow up properly, and worth skipping when you do not. The money is rarely lost in the booth rental. It is lost in the two weeks afterward.

Selling through distributors and manufacturers’ representatives#

If you sell through distributors or manufacturers’ representatives, marketing is still your job, with a different target. The work is to create demand that reaches your channel already asking for you by name, which the research on industrial channels calls a pull strategy, and you measure it in specifications written instead of leads collected.

A study of 195 purchasing managers found that when a component supplier created demand beyond the channel, as well as selling into it, its position improved, and that the right balance between the two shifts with how easily the customer could make the component themselves[43]. It covers one component in one industry in China with state-owned customers, so take the idea and leave the exact figures[43].

In practice, a channel partner has its own strategy and reads everything you publish through it. A distributor sees a capability page ending in a where-to-buy link as support, and the same page ending in a quote form as the first step toward being cut out. Which one you build is a decision about the relationship more than about the website.

How to publish without cutting out your own channel

• Publish the technical data and send the sale to your channel. Specifications, drawings, compatibility and part data belong on your site, with the where-to-buy link right next to them.
• Name your distributors on your own site. It costs nothing, it strengthens the relationship, and it answers the one buyer question your channel cannot answer for you, which is whether you are real.
• Give the channel the content. Photographs, cut sheets and short videos a distributor can use are worth more to them than a co-op advertising budget, and you make them once and use them twice.
• Count specifications as your result. If your name appears on a print, you won something a lead count will never show. Ask the distributor which jobs came in already naming you.
• Decide in advance whether you will ever sell direct. Whether you will quote direct, and when, is a policy your channel would rather hear once than discover.

The marketing that fills your open roles#

For most manufacturers, the highest return on the whole marketing program is a new hire, ahead of a new customer. The same website, the same photographs of the floor and the same social presence do both jobs, and the labor market for skilled manufacturing work is tight enough that a filled position is often worth more than an extra quote request.

If you have lost a good machinist to the shop down the road, you know candidates look you up too. US manufacturing had 580,000 job openings in July 2026, on a monthly survey of about 21,000 establishments[44]. That figure is preliminary and an unusually sharp jump from the previous month, so the twelve-month average nearer 462,000 is the safer number to plan with[44]. Modeling by Deloitte with The Manufacturing Institute suggests the sector could need 3.8 million workers between 2024 and 2033, with roughly 1.9 million positions going unfilled if the gap is not closed[45].

That 1.9 million is a modeled scenario that assumes nothing changes, and the report presents it that way instead of as a forecast[45]. Even discounted heavily, it describes a market where people, more than orders, limit how fast a plant can grow.

Where applicants actually come from, ranked

If a hire is the highest return in this guide, hiring deserves the same treatment as customer marketing. The order below is ours, from doing this work with plants.

What a careers page has to carry

• State the shift patterns and the actual hours. It is the first question every applicant has, and the one most rarely answered before an interview.
• Publish the pay range. It can feel uncomfortable, it is increasingly expected, and it is the fastest way to stop wasting everyone’s time.
• Describe the work in the words a machinist would use. Name the machines, the materials, the batch sizes, and what a day looks like.
• Show photographs of the actual floor, including the parts of it that are not new. Applicants have worked in plants before and can tell when they are being shown a different building.
• Name the person they would report to. It turns an anonymous employer into a person.
• Offer a way to apply that works on a phone, without an account, a portal or a required resume upload.

The overlap with customer marketing is close to total. A buyer deciding whether you are a real company and an applicant deciding whether to leave a stable job are looking for the same evidence, and both will read the careers page.

What to measure when the sales cycle runs nine months#

Count requests for quote, the value quoted, and the win rate, each by source. Traffic charts move months before revenue does, and a program judged on visits will be steered toward the cheapest possible visitor. The one number worth reporting to an owner is the quoted dollars that came from marketing, followed by how many of those quotes were won.

If you have ever been handed a monthly marketing report and could not tell whether anything was working, you are not alone. This is where most industrial marketing reporting goes wrong, and it is easy to fix. A report of sessions, bounce rate and impressions describes the activity and says nothing about the result. Nobody on a shop floor is paid for impressions.

The difficulty is well documented. Among manufacturing marketers, 64% name attributing return on investment as a measurement challenge, 53% struggle to connect content to business goals and 48% report unclear key performance indicators[36]. Those three describe one problem. A long buying cycle makes the link between an action and an order hard to see, so teams fall back on the numbers that move every week.

The measurement setup, in the order to install it

A plant does not need a pile of marketing software. It needs four things set up correctly once, and a fifth that is a simple sheet.

1. Verify your domain in Google Search Console. It is free, and it is the only place that shows the actual searches people used to reach you. For this purpose it is worth more than any analytics package.
2. Install analytics with a conversion set up on the quote form. Setting up that conversion matters far more than which package you choose. Without it, the package measures nothing you can act on.
3. Put a separate phone number on the website, so you can tell calls from the site apart from calls that came from a catalog printed in 2009. This one change settles most arguments about whether marketing is working.
4. Send quote requests to a shared inbox. Requests that land in one person's mailbox are invisible the week that person is on vacation, and impossible to count in any month.
5. Keep a sheet with one row per quote request, showing the date, the company, what they wanted, where they came from, the quoted value, and whether you won or lost it. A spreadsheet is all you need, and it will do better than a CRM nobody keeps up.

One habit is worth more than any dashboard. Every week, ask the person who handles quotes whether the inquiries arriving are worth their time. That answer comes months before the revenue figures, and it is usually right.

What a manufacturing marketing program costs#

A realistic program for a manufacturer under $50 million in revenue runs between $500 and $4,000 a month, plus a one-time website build. Our own published prices sit inside that band: a website at a fixed $6,000 one time, search work at $675 a month, and social media management at $1,900 a month. If you have sat through vague quotes from agencies that will not name a figure, they are usually pricing off your revenue instead of off the work.

A range is more useful than a single number, because what you should spend depends mostly on how much of the writing and photography your own team can do. The capability text is best written by somebody who knows the machines, and that person already works for you.

A range is a starting point, and you still need a plan. Working backward from the revenue you need is more useful, and the jobs, quotes and traffic a target implies returns a plain verdict on whether the plan is plausible before you spend anything. It needs one figure most manufacturers have never calculated, which is what a customer is worth across the whole relationship rather than on the first order.

Set the budget from what one job is worth

The common advice is to spend a percentage of revenue on marketing. For a job shop it is close to useless, because two shops with identical revenue can have average job values ten times apart, and the figure that decides everything is what one more quote request is worth to you.

Work it the other way round. Multiply your win rate by your average job value by your gross margin. The result is the gross profit one quote request carries before it is won or lost. Divide a proposed monthly spend by that figure, and you have the number of extra quote requests a month the spend has to bring in to break even. If that number sounds out of reach for your market, the budget is wrong.

What to cut first when the budget gets cut

Budgets get cut mid-year, and the order you cut in matters more than the amount. Cut from the bottom of this list upward, and the program gets through a lean quarter with the work that builds over time still intact.

1. Keep the website accurate. It costs almost nothing, and everything else depends on it. A wrong certificate scope or a dead phone number undoes the whole program.
2. Keep the Google Business Profile current. It is free, and it is the one listing that brings in calls without any further spend.
3. Keep publishing capability pages, even slowly. One page a month still builds on itself, and zero pages a month does not.
4. Cut paid advertising next. It stops the day you stop paying, which is exactly what makes it the right thing to pause.
5. Cut social media production before search work. Social resets to zero when you stop; a page that ranks keeps working.
6. Cut the redesign. A redesign is the most expensive thing on most marketing plans and the least likely to change the number of quote requests.

The costs that get left off the plan

• Your team's time is a real cost. Two hours a week of an operations manager's time costs money, and the program fails without it.
• Photography is usually forgotten. One day of a photographer on your floor, once, is worth more than a year of stock images.
• The domain and the accounts need an owner. They cost little and are easy to lose, and they become very expensive the moment you find out somebody else holds them. If changing a phone number on your site has ever taken a week and an invoice, you know the feeling.
• Quoting takes capacity. A program that works brings in more quote requests, and every quote takes time to prepare. Success has a cost, and it is worth planning for.

Who should run it: in house, agency, or both#

Somebody inside the company has to own marketing, even at a quarter of their time, because only an insider knows the machines, the schedule and which jobs are worth chasing. Outside help is most useful for the technical build, the structure of the pages and the ongoing search work. A program with no internal owner produces content nobody at the plant recognizes.

We know the website usually lands on whoever has a spare hour, which is nobody. The good news is that the person who owns it inside your company does not need marketing experience. They need the authority to get a photograph taken, access to the person who writes quotes, and about two hours a week, of which roughly thirty minutes is the owner’s. An operations manager or an engineer who writes clearly is usually a better choice than a junior hire with a marketing degree and no access.

If you are choosing between hiring that person and buying the work, run the numbers before you have the argument. A salary priced against a retainer and a freelancer includes payroll taxes, benefits, equipment and the months before a new hire is productive. On cost per hour, its result does not favor us.

The internal owner's actual job, written out

The role fails when someone is told to own marketing and nobody turns that into specific, recurring work. Written out as a list, it takes about two hours a week, and any capable operations person can do it.

• Each week, they take five photographs on the floor. They also send the week's quote requests, and where they came from, to whoever is tracking them, and answer any technical question the writer has asked.
• Each month, they read anything about to be published. They check every number in it, because one wrong tolerance on a capability page costs more credibility than the page earns.
• Each quarter, they sit down with whoever prices the work. They ask whether the inquiries arriving are worth that person's time, and update anything that has changed, such as a new machine, a new certificate or a capability you no longer offer.
• Once a year, they confirm who controls the accounts. That means the domain registration, the DNS, the analytics account and the Google Business Profile. This is the gap manufacturers most often discover when they want to change agencies.

Three models, and who each one suits

What to ask an agency before signing anything

1. Who owns the website when this ends, and what do I have to pay to leave? Ask for the answer in writing, with the source code, domain, analytics and hosting all named.
2. What will you publish in the first ninety days, specifically? A plan that cannot name the pages is a retainer looking for a purpose.
3. Who writes the technical content, and how will they get it right? If the answer does not involve interviewing whoever quotes your work, the pages will be generic.
4. What will you report, and will it include quote requests? A report built on sessions and impressions cannot tell you whether to continue.
5. What are you not good at? A direct answer here tells you more than anything else before a contract is signed.

Two comparisons are worth reading before you choose: agency against freelancer and in-house against outside help. Both compare kinds of help, and both include the cases where the right answer is to hire nobody.

A 90 day sequence you can start on Monday#

Ninety days is enough to fix the website, publish the pages that carry your capability, claim the listings and start counting quote requests properly. Ranking for a competitive category term takes longer than that, and any plan that promises otherwise is describing paid traffic. The sequence below assumes one person inside your company giving it a few hours a week, and each step makes the next one cheaper.

1. Weeks 1 and 2: write down who buys and what they type. List the three or four job titles that appear on a purchase order for your work, and the exact phrases each of them would type into a search box. Use the words on your own drawings and quotes, because those are the words your buyers use. This list becomes the plan for your pages.
2. Weeks 2 and 3: fix what the site says about capability. Put your processes, materials, tolerances, machine envelopes and certifications on the site as readable text. If your capability statement lives only inside a PDF or an image, write it out as text on a page, where buyers and search engines can both read it.
3. Week 4: shorten the path to a request for quote. Add a quote form that buyers can reach from every page in one click, and let it accept a drawing file, a quantity and a target date. Put a named person and a phone number next to it, so the buyer knows who will read it.
4. Weeks 5 to 8: build one page per thing you sell. Give each process, material and industry you serve its own page with its own address. One page that covers eight capabilities ranks for none of them.
5. Weeks 6 to 9: publish two case studies with numbers in them. Take two completed jobs and write down the part, the problem, the process, the tolerance you held and the measured outcome. Get written permission from the customer for anything they could consider confidential.
6. Weeks 9 to 12: claim the listings and start measuring quotes. Claim and complete your Google Business Profile, and make the company name, address and phone number identical everywhere they appear. Then start counting requests for quote by where they came from, since visits alone tell you very little.

Who does which part

The sequence stalls in the same place every time: a step that needs somebody inside the company gets handed to somebody outside it. Agree who does what before you start.

Months four to twelve, where the return actually shows up

The first ninety days build the foundation. The next nine months are when it starts paying, and the work in them is deliberately unglamorous.

• In months 4 to 6, publish two pages a month. Watch Search Console for the searches that are starting to show your site, and write the page each one suggests. This is the cheapest content research there is, and almost nobody does it.
• In months 6 to 9, add the second and third case study. Ask the assistants your buyers use whether they name you yet, and write down who they name instead.
• In months 9 to 12, run the first full review. Compare quote requests by source against the same period last year. This is the first comparison with enough data behind it to mean anything.
• Throughout the year, keep the small habits going. Take photographs every week and keep the quote log current. Both take minutes, and both are the first things to slip.

Rankings on the narrower terms usually begin moving somewhere between month three and month six, and the quote requests follow a few weeks later. Anyone promising faster results than that on competitive terms is describing paid traffic.

The failure modes we see most often#

The same eight mistakes account for most wasted industrial marketing budget: buying traffic before fixing the site, writing for the engineer alone, hiding capability inside PDFs, chasing category keywords, gating the material that would have won the shortlist, redesigning instead of publishing, judging the program on sessions, and stopping at month four.

The last one does the most damage. Industrial search work builds slowly and then holds, so a program canceled at month four has paid the whole cost and collected none of the return. If you agree the time frame before the first invoice, you will not have to make that decision in a month when the chart looks flat.

How to tell which one is yours

Each mistake has a symptom that shows up before the invoice does. Go through these with whoever answers the phone and whoever writes the quotes, because between them they can check the whole list in twenty minutes.

• Ask what people say when they call. If callers open with a question your website already answers, the site is not answering it where they looked.
• Open your own site on a phone, on cellular, standing on the floor. Time how long it takes to find the tightest tolerance you hold. If you cannot do it in thirty seconds, a stranger will not manage it at all.
• Search for your own capability without your company name. Use the words a buyer would use instead of the words on your sign. If you are not on the first page, you are not on their list.
• Count the quote requests that arrived last month and where they came from. If nobody can answer that from memory or from a sheet, the program has no measurement and cannot be judged.
• Read the last five quotes you lost. If they were lost on price, you have a sales problem. If they were lost because you were never asked, you have a marketing problem.

Four more that deserve more attention than they get

1. Letting the domain expire, or moving it badly. A move to a new site done without redirects can wipe out years of rankings in an afternoon. Find out today who owns the domain registration and when it renews, well before the day it lapses.
2. Publishing a phone number nobody answers. An industrial buyer with a stopped line calls once. A number that goes to voicemail during a shift is worse than no number, because it turns an opportunity into a specific bad impression.
3. Treating the careers page as HR's problem. It is often the second most visited page on a manufacturer's site, and the same people who read the capability pages read it too. A neglected one tells a buyer something about how the plant is run.
4. Letting the certificate scope go stale. A certification page showing an expired certificate is worse than a page with none, because a quality reviewer reads it as a sign that your document control has slipped.

The evidence base, and how to read a marketing statistic#

Every figure in this guide comes from one of 46 named sources, listed in full at the foot of the page with its sample, its date and its limitations. 10 are peer reviewed, 11 are surveys with published methods, 8 are official government or platform sources, 9 are industry reports and 8 are analyses. Where a figure could not be verified to a primary source, the reference says so.

Industrial marketing has a statistics problem. Figures circulate for years with no traceable origin, get rounded differently by each republisher, and end up justifying decisions their original authors never made. The 57% claim examined earlier is the most famous example, and plenty of others behave the same way. So this guide names its sources and the limits of each one, so you can judge every figure for yourself.

Four questions to ask of any marketing statistic

1. Who was asked, and how many? A finding drawn from 104 manufacturing marketers[36] is worth having, and it is still a small sample. Half of it worked at companies with a thousand or more employees, far larger than most of the manufacturers this guide is written for.
2. Were they the right population? The Pew finding on AI summaries is drawn from a general US adult panel[20], and engineers behave measurably differently[14]. Using the general figure to plan an industrial site would lead you to the wrong conclusion about whether the click still arrives.
3. How old is it? The widely quoted buying group figures come from a 2017 survey[6]. That is old enough to treat as a general picture instead of a current measurement.
4. Can you reach the primary source? Several of the most quoted figures in this field sit behind a form or a paywall, and one publisher's pages could not be opened at all when this guide was checked. Where that happened, the reference says so.

Questions manufacturers ask about all of this#

Here are eleven questions, each answered on its own. They cover the four things owners ask before committing money: how long it takes before anything arrives, what it costs at your job value, what your own people have to do every week, and how you will know whether it worked.

How long before digital marketing produces a request for quote?

Fixing an existing site can bring in quote requests within weeks, because the visitors you already get start turning into requests. Ranking for specific process and material searches usually takes three to six months, and competitive category searches take longer. If you judge a program at week six, you are judging it before it has had the chance to produce anything.

Is SEO still worth it when AI assistants answer the question directly?

Yes, because the assistants read the same pages Google does. When someone asks an assistant for a supplier, it reads indexed pages, third-party listings and reviews, then names a shortlist. If your website has no capability text, you are missing from both the search results and the assistant’s answer, and the work that fixes one fixes the other.

We sell through distributors and reps. Does any of this apply?

It applies, with a different goal. When you sell through channel partners, being found in search creates specification pull: the engineer finds your data, names your part on the drawing, and the order reaches you through the distributor. Publish your part data, drawings and compatibility information so the engineer specifying the part can find you in the first place.

How much of this can we do without hiring anyone?

Someone who knows the work can handle the capability text, the certifications, the case studies and the Google Business Profile in-house, and those four cover most of what there is to gain. Outside help is most useful for the technical build, the structure of the pages and the ongoing search work.

Do we need a blog?

You need pages that answer the questions buyers ask, which is a different thing from a dated blog. A permanent page about a process, a tolerance or a material keeps earning quote requests for years, while nine posts about company news earn nothing. If a topic will still be true in three years, give it a permanent page of its own.

How do we market work we are not allowed to show?

Publish the capability and leave out the customer. Describe the process, the material, the tolerance and the volume, and keep the part and the customer unnamed. Say plainly that you work under confidentiality and will discuss specifics under an agreement, because to the right buyer that sentence is a credential in itself. Where the work is export controlled, name the regimes you operate under, since buyers looking for that capability search for exactly those words.

Our average job is small. Does any of this pay for itself?

You can work it out in a few minutes. Multiply your win rate by your average job value by your gross margin to get the gross profit one quote request carries, then divide a proposed monthly spend by that figure. The answer is the number of extra quote requests a month you need to break even. At a $2,000 average job with a 25% win rate and a 30% margin, one quote request carries $150, so spending $1,500 a month needs ten extra quote requests a month. If that number sounds out of reach for your market, spend less, or put the money somewhere else.

What should we cut first if the budget gets cut?

Cut in this order: the redesign first, then social media production, then paid advertising. Keep the website accurate, keep the Google Business Profile current, and keep publishing capability pages, even if it is only one a month. The work that builds on itself over time should be the last to go, because stopping it throws away the months you have already put in as well as the money.

How do we know whether a quote request came from marketing?

You will never know it perfectly, and any tool that promises to tell you exactly is claiming more than it can see. Use three imperfect measures together: a separate phone number on the website, a conversion set up on the quote form in your analytics, and one question on that form asking how the buyer heard about you. That last one is the least precise of the three and usually the most useful.

What does the person inside our company actually have to do?

About two hours a week. Each week, they take five photographs on the floor and pass on the quote requests and where they came from. Each month, they read anything about to be published and check the numbers in it. Each quarter, they ask whoever prices the work whether the inquiries are worth their time, and update anything that has changed. Once a year, they confirm who controls the domain, the DNS, the analytics and the Google Business Profile.

What does a marketing program cost for a manufacturer of our size?

Our own prices are fixed and published: a website build is $6,000 one time, search work is $675 a month, and social media management is $1,900 a month. For a manufacturer under $50 million in revenue, a useful budget is $500 to $4,000 a month plus the build, and the split depends on how much of the writing your own team can do.

Where to go next

• Every common buyer question is answered in full on its own page.
• The RFQ readiness scorecard scores your site on twenty weighted checks.
• The marketing calculator works out cost per quote request, payback and break-even.
• Every price we charge is published, with what each one includes.
• The PLI case study covers 5 brand sites for one industrial group.
• What RFQ-first means explains why we organize the work around quote requests instead of traffic.