AEO for manufacturers

The Engineering Knowledge Programme

Your buyers used to phone three suppliers they already knew. Now a growing number of them open ChatGPT and ask who can hold ±0.02mm on a 316 stainless part, in small batches, in the UK. Something answers. It probably isn't you.

Answer engine optimisation for subcontract engineering, fabrication and industrial manufacturers. Built and run by Simon Young, Doncaster and South Yorkshire, working nationally.

Book a strategy call What AEO is, in general

Last updated 12 August 2026

What changed

Search didn't die. The click did.

Google's AI Overviews now appear on roughly half of all queries. When one appears, click-through on the top result falls by somewhere between a third and two thirds depending on query type. Around six in ten searches now end without anyone visiting a website at all. Rankings hold. Impressions hold. Sessions collapse.

In manufacturing this shows up differently to most sectors, and more dangerously. Your website was never a high-traffic asset. It was a credibility check, the thing a buyer looked at once, halfway down a shortlist they'd already built somewhere else. So when enquiries thin out, nobody blames the website. They blame the market, the framework agreement, reshoring, the Germans, the Chinese.

But the shortlist is now being built somewhere new. It's being built inside an answer, by a machine that has read whatever it could find about who does what, to what standard, in what volume, in what part of the country. If you haven't put that knowledge where it can find it, it's using someone else's.

And the traffic hasn't moved channel. In sector after sector the clicks stopped rather than migrated, AI referral traffic is still a low single-digit share of all web visits. "We'll pick up our AI traffic instead" is not a plan. The plan is to be inside the answer, because for a growing share of your buyers the answer is the shortlist. They never arrive at a website to be converted. They arrive at an RFQ, and you're either on it or you aren't.

Knowledge is the asset now

For twenty years, search was a ranking game. You competed for a position on a list, and the list reset every time Google changed its mind. Whatever you built was rented.

Answer engines don't work like that. They don't rank, they decide who to trust and then repeat what they've learned. That decision is built on accumulated, corroborated knowledge: what exists about you, what other sources say about you, how consistently it all agrees, and whether you're the source that actually answered the question.

For a manufacturer that's an unusually good deal, because you are sitting on more genuinely citable knowledge than almost any other kind of business. Tolerances. Material behaviour. Process limits. Lead-time realities. Design-for-manufacture advice you give away on the phone twenty times a week. None of it is published. All of it is exactly the specific, factual, structured, attributable content answer engines prefer to cite.

It compounds

Every answer you publish and every third-party source that corroborates it makes the next citation easier to win. Traditional SEO decayed the moment you stopped. This accrues.

It's defensible

Once a model is confident you're the authority on small-batch aerospace-approved turning in Yorkshire, a competitor can't outbid you for that position, because there's nothing to bid on. They have to build a deeper, better-corroborated knowledge base than yours, from behind, while you keep building. That's a moat, not a ranking.

It's empty right now

This is the part I'd take seriously if I were you. Ask an assistant to recommend a fabricator, a machine shop or a specialist finisher in your region and you'll typically get two directory aggregators, whichever firm happens to have an old trade listing, and a shrug. There is no incumbent in industrial manufacturing. In most capability-and-geography combinations in the UK, the position is completely unclaimed.

The window

The land-grab period in any new discovery channel runs eighteen months to two years, and this one started roughly a year ago.

What makes this one unusual is that being early doesn't just help, it entrenches. Because models weight corroborated, long-standing, consistently-agreeing sources, the firm that establishes itself as the answer becomes the reference point later entrants have to dislodge. Being first on Google meant being first until the next algorithm update. Being first here means being the thing your competitor's content gets measured against.

Manufacturing also has a specific advantage and a specific risk. The advantage is that your sector is slow to adopt marketing technology, so the window here is wider than in, say, dentistry or law. The risk is that when one firm in your niche does move, they'll move into an empty room and take all of it.

Why me

I've been writing about answer engine optimisation since October 2019, more than three years before ChatGPT existed. I followed it that November with a piece called "The Death of the SEO Industry." At the time it read as a provocation.

That's not a vanity credential. It matters because it means I built this thinking from first principles, from how machines decide what to trust and repeat, rather than reverse-engineering it from a blog post in 2024, like most of the "AEO agencies" that appeared eighteen months ago. When you're buying help in a category this young, the honest question to ask any supplier is: what were you doing before this was a category? Most can't answer it. I can, and the dated receipts are published.

A research-and-publish pipeline that makes volume possible. Mining the real questions your buyers and their engineers ask, from search data, People Also Ask, trade forums, technical communities and your own RFQ inbox, then researching and building genuinely answer-shaped pages against them, quality-scoring each one, and getting them indexed fast.
Multi-engine citation tracking with evidence. Your named prompts run repeatedly across ChatGPT, Gemini, Claude, Perplexity, Copilot and Google's AI surfaces, with screenshots. You see where you appear, who appears instead of you, and what moved.
Entity and citation architecture. Schema, named-engineer authorship, accreditation markup, third-party corroboration, directory and profile consistency. All the unglamorous scaffolding that decides whether a model is confident enough to say your name out loud.
I speak shop-floor. I'll sit in your office, walk the floor, and pull the questions out of your estimator's head. Nothing gets published that your technical director wouldn't say out loud to a customer.
Plain English, always. If I can't explain what I did and why it worked to a non-technical MD over a pint, I haven't done it properly.

One honest warning

There's a wave of "AEO" out there that's a rebranded SEO retainer with a new sticker. The tells are easy.

1.They conflate AEO and GEO and can't tell you the difference.
2.They promise multi-LLM citation tracking but can't produce screenshot evidence against a defined prompt set.
3.They treat schema as a one-off setup.
4.They tell you third-party authority doesn't matter, usually because they can't do it.
5.Manufacturing-specific: they'll write your technical content without ever speaking to an engineer. Ask to see a sample. If it could describe any machine shop in Britain, it will get you cited by nobody.

Ask me those five questions. Then ask the next agency the same five.

The question map

Two audiences, two entirely different question sets

This is what makes an engineering programme different from a generic AEO retainer. You have two audiences asking two entirely different question sets, and almost nobody optimises for either.

Buyer-side: procurement, purchasing and the people with the budget
Type	Example prompts
Selection	“Who does small-batch CNC machining in South Yorkshire?” · “UK sheet metal fabricators for stainless enclosures” · “AS9100 approved subcontract machinists near me”
Capability	“Can anyone in the UK 5-axis machine Inconel in low volume?” · “Who does laser cutting and forming under one roof?” · “Which UK suppliers hold ISO 13485 for machined components?”
Commercial	“What is a reasonable lead time for machined parts?” · “How much does a fabrication prototype cost?” · “Typical MOQ for injection moulded parts UK”
Risk	“How do I qualify a new subcontract supplier?” · “What should be in a supply agreement for machined parts?” · “How do I check a supplier's quality accreditation?”

Engineer-side: the technical questions that build your authority
Type	Example prompts
Process	“What tolerance can 5-axis machining realistically hold?” · “Laser cutting vs waterjet for 20mm stainless” · “When should I choose casting over machining?”
Material	“316 vs 304 stainless for marine use” · “Best aluminium grade for anodising” · “Will this material work-harden during machining?”
Design	“Design for manufacture rules for sheet metal bending” · “Minimum wall thickness for aluminium castings” · “How do I reduce machining cost on this part?”
Standards	“What does AS9100 actually require?” · “PPAP vs FAIR, what's the difference?” · “What is a material certificate 3.1?”

Three things fall out of this map, and all three are commercial

The engineer-side is cheap, enormous and completely uncontested. These are factual, structured, updatable technical questions with real volume, and virtually no UK manufacturer has published serious answers to them. Own the process and material answers for your capability and you become the cited technical authority, which bleeds directly into the buyer-side selection answers. You win the purchasing manager by teaching the design engineer.
"Is [your company] any good?" is already being answered about you. By something, based on whatever it could find, an old directory entry, a Companies House filing, a five-year-old news mention, a review you've never seen. Most manufacturers have never once checked what that answer says. It's the first thing the audit looks at, and it's occasionally alarming.
The money sits at capability × material × sector × geography. Not "CNC machining UK", which is contested and worthless. But "small-batch 5-axis machining, Inconel, aerospace approved, Yorkshire" is defensible, has real order value behind every conversion, and in most cases literally nobody has built the knowledge to own it.

That grid is what the levels are actually buying: how many of those squares you claim, and how hard.

The levels

The three levels

All prices per month, excluding VAT. Three-month minimum, then rolling with 30 days' notice. A twelve-month commitment takes 10% off.

	Visibility	Recommendation	Dominate
Monthly	£1,250	£1,995	£3,495
The promise	You exist to the machines	You're on the shortlist	You are the answer
Best for	Single-site shops, one core capability	Multi-capability or multi-site manufacturers	Sector specialists taking or defending a niche
Knowledge assets / month	8	15	30
Tracked prompts	30–40	100–120	250+
Engines	ChatGPT, Google AI, Perplexity	Plus Gemini, Claude, Copilot	All, monitored weekly
Coverage	1 capability × 1 region	Up to 3 capabilities × 3 sectors	Unlimited in agreed scope
Third-party citation building	Foundation only	Active programme	Active plus editorial placement
Original data asset	·	·	Annual lead-time & capacity index
Area exclusivity	·	·	Yes
Reporting	Monthly	Fortnightly	Weekly plus live dashboard
Simon's time	Monthly review call	Fortnightly working session	Weekly, plus quarterly on-site

Onboarding and build: £995 one-off. That covers entity architecture, schema build, accreditation markup, baseline measurement and question-bank construction. Waived on a twelve-month commitment.

Front door: the AI Visibility Audit, £495 one-off. Credited in full against your first month if you start within 30 days.

Level 1, Visibility

£1,250 / month

“Right now, when a buyer asks an AI who can make their part, you don't come up. This fixes that.”

The problem it solves: you're invisible. Not badly ranked. Absent. The models have no confident basis on which to say your name, because nothing they can read tells them what you actually do.

Baseline and question bank. 30–40 high-intent prompts across buyer-side and engineer-side questions, agreed with you, tracked monthly across ChatGPT, Google's AI surfaces and Perplexity with screenshot evidence.
Entity foundation. Organisation, LocalBusiness, Service and FAQ schema built and maintained. Accreditations, certifications and approvals marked up so they're machine-readable facts rather than a logo strip. Named-engineer authorship so your technical people are entities, not anonymous bylines. Name, address and profile consistency corrected across the trade directories and databases the models actually read.
8 knowledge assets per month, built from your question bank: capability pages, material guides, tolerance and process explainers, accreditation answers. Answer-shaped, written in your voice, checked by a human who understands what a datum is, never generic.
Google Business Profile discipline. Weekly activity, category correctness, review prompting cadence. Still one of the strongest local signals and still neglected by nearly every manufacturer in the country.
Monthly report and review call. What we published, what moved, where you appear now, and who appears in your place where you don't.

Not included: third-party editorial placement, competitor displacement work, multi-capability scale, weekly monitoring.

Honest expectation: first movement in citation rates typically shows at 30–60 days as models refresh. Meaningful shortlist presence in your core capability by month four to six.

Level 2, Recommendation

£1,995 / month

“Being findable isn't the game. Being named is. This gets you onto the shortlist of three that goes out to RFQ.”

The problem it solves: you turn up occasionally, in passing, usually below a directory aggregator or a national group. You're in the room but you're not the recommendation.

Everything in Visibility, plus:

100–120 prompts across six engines, tracked fortnightly with evidence: ChatGPT, Gemini, Claude, Perplexity, Copilot and Google AI Overviews / AI Mode.
Up to three capabilities across three sectors. This is where the capability × material × sector grid gets built properly, with an architecture that doesn't cannibalise itself.
15 knowledge assets per month. Enough volume to own a question category rather than dabble in it, including the full engineer-side process and material set, which is where the compounding technical authority comes from.
Active third-party citation building. Trade bodies and associations, sector directories, regional business and manufacturing press, supply-chain listings, customer and partner mentions, framework and approved-supplier listings. This is the majority of what makes a model confident enough to recommend you, and it's the line item cheap providers quietly skip.
Competitor citation analysis. Every fortnight: who's being named instead of you, on which prompts, and what they have that you don't. Then we go and get it.
Named-engineer authority. Your technical director and senior engineers become citable experts in their process, through bios, authored answers and quoted commentary. Models cite people as well as brands, and in engineering the person carries the credibility.
Fortnightly working session. Not a status update, a working half-hour on what's moving.

Honest expectation: consistent presence in shortlist answers for your primary capability by month four to six. Measurable displacement of at least one incumbent name by month six to nine.

Level 3, Dominate

£3,495 / month

“When a buyer anywhere asks who can make that part, to that standard, in that volume, yours is the name that comes back. Every engine. Every time.”

The problem it solves: you're on the shortlist and you want to be the shortlist. Or you're a genuine specialist with a window to close before someone in your niche notices it's open.

Everything in Recommendation, plus:

30 fully researched pieces per month. Every one built to fill a specific, identified knowledge gap in your process and sector: a question your market is asking that nobody has properly answered. Researched, not spun. Sourced, structured, attributed to a named engineer, and published to compound. Three hundred and sixty pieces a year, each one a brick in a technical knowledge base nobody else has.
250+ prompts, weekly monitoring, live dashboard. You see movement as it happens, not a month later.
Unlimited capabilities and sectors within agreed scope. Full grid coverage across every process, every material, every served industry, both audiences.
Editorial placement programme. Named features and technical commentary in trade press, conference and exhibition coverage, industry report contributions, podcast appearances. High-corroboration third-party sources that models weight heavily and competitors can't shortcut.
Your own original data asset. A regional lead-time and capacity index, or a materials and machining cost tracker, published under your name and updated quarterly. Original data is the most citable asset class there is, it earns links, press and AI citations simultaneously, and it turns you from a source into the source.
Competitor displacement campaigns. Targeted work on the specific prompts where a named rival currently owns the answer.
Area exclusivity. While you're on Dominate, I won't take a directly competing manufacturer in the same capability and region. That's not a sales tactic, it's the only way the work stays honest, and it's why this level is capped.
Weekly reporting, plus a quarterly on-site. I come to you. Half a day on the floor and with your estimators, pulling the real questions off the desk and into the pipeline.

Honest expectation: category ownership in a defined capability and region within six to nine months. This is also the level where results depend most on you. Floor access, engineer time and permission to publish real technical detail are what make it work.

The audit, £495

Before any retainer, and available entirely on its own.

Your current citation position across all major engines on 25 defined prompts, with screenshots
What the machines currently say about your capability, capacity and accreditations, including where they're wrong
Which competitors are named in your place, and why
Your entity, schema and accreditation gaps, listed and prioritised
A twelve-month roadmap: the capability-and-material squares worth owning, in the order worth owning them
Delivered in five working days, in plain English, yours to keep whether you hire me or not

Credited in full against your first month if you proceed within 30 days.

Some manufacturers will read the audit and conclude the opportunity in their niche is too thin to justify a retainer. I'd rather they found that out for £495 than for £12,000.

The ROI conversation

Manufacturing maths is less tidy than a recruitment fee, so let's use contribution rather than revenue and keep it conservative.

Take one recovered subcontract order worth £25,000 at a 30% gross margin. That's £7,500 of contribution from a single job you'd otherwise never have been asked to quote.

Level	Monthly	One £25k order at 30% margin covers
Visibility	£1,250	Six months
Recommendation	£1,995	Nearly four months
Dominate	£3,495	Just over two months

You're not comparing this to your marketing budget. You're comparing it to one job. The question is whether being the named answer for your capability is worth one extra quote request a quarter.

And that's the pessimistic version, because subcontract work rarely arrives as a single order. A new customer who finds you this way and stays is a repeat account, and one repeat account funds the programme permanently.

Questions manufacturers ask

Isn't AEO just SEO with a new name?

No, and the difference is measurable. SEO gets you a link on a page a shrinking number of people click. This gets you named inside the answer, which is increasingly the entire interaction. They share some plumbing, they don't share an outcome. Worth saying plainly: we were writing about answer engine optimisation in 2019, it isn't a rebrand.

Our customers are long-standing accounts won by relationships. Does this matter?

It matters for the accounts you haven't won yet, and for the day a long-standing buyer retires and their replacement builds a new supplier list from scratch. That replacement will not phone round. They'll ask a machine who's good, and take the answer seriously.

We don't want to publish our technical know-how.

Nothing gets published that you don't approve, and there's a clear line between the knowledge that wins you trust and the knowledge that wins you jobs. Explaining what tolerance is achievable on a process is not giving away your setup sheets. The firms that publish generously are the ones models trust, and your competitors are, right now, being asked those same questions on the phone and answering them for free anyway.

How do I know it's working?

Screenshots. The same defined prompts, every cycle, every engine, with evidence of where you appear and who appears instead of you. If a supplier can't show you that, they aren't measuring, they're describing.

Can't our marketing person just do this with ChatGPT?

They can generate pages. What they can't easily do is question mining at scale, entity and accreditation architecture, third-party corroboration, which is most of what actually moves a model, and the measurement discipline to know whether any of it worked. Volume of generic content is currently the fastest way to get filtered out.

Why now? Can't this wait until the order book quietens down?

It can. It'll cost more and take longer. Right now the position in your niche is empty and you're building against nobody. In eighteen months you'll be building against whoever took it, and models weight established, corroborated sources.

How much does AEO for a manufacturer cost?

Three levels, per month excluding VAT: Visibility £1,250, Recommendation £1,995, Dominate £3,495. Onboarding and build is £995 one-off, waived on a twelve-month commitment. The front door is a £495 AI Visibility Audit, credited in full against your first month if you start within 30 days.

I turn down more firms than I take on. If your niche is too crowded for this to pay back, or you're not going to give me the engineer time it needs, I'll tell you and walk away. I'd rather have five clients who own their categories than fifty who move sideways.

Claim your capability

Start with the £495 audit and see exactly what the machines say about your company today, who they name instead of you, and which capability-and-material questions are worth owning first. Credited in full against your first month.

Book a strategy call How we measure

Not your sector? See the same programme built for recruiters, dentists, financial advisers, IT support providers and care homes. Or read more on how AI chooses who to cite.