ChatGPT
Browses live, and leans on brand familiarity
Consistent descriptions and third party coverage move it most
03 Get named when the model answers.
Generative Engine Optimisation for AI visibility
When somebody asks ChatGPT for a plumber in Wolverhampton or a restoration firm near Coventry, a handful of businesses get named. Generative Engine Optimisation is the work of being one of them.
Asked “who should I use for this in Wolverhampton”
- Your own site Weighted lowest
- Your Business Profile Checked for agreement
- Everyone else writing about you Weighted highest
The answer names three
- A competitor
- Your business
- A national chain
A recommendation you were not in the room for
Search used to be a list you could inspect. A generative answer is a recommendation, made in a conversation you cannot see, and the shortlist is already narrowed by the time your buyer reads it. Three names get mentioned. The fourth best firm in the county does not get a sentence.
Models do not rank pages, they assemble an answer from what they can find and verify about an entity. That changes what the work is. Being described consistently across the web matters more than a title tag. Being written about by somebody else matters more than writing about yourself. Having a clear, machine readable account of what you do, where you do it and what you are qualified for matters more than keyword density ever did.
This is the newest service we sell and we will be straight about it: the measurement is less mature than SEO reporting, the platforms change their behaviour without notice, and nobody can promise a position. What can be done is make you the easiest firm in your sector to describe correctly.
What Generative Engine Optimisation involves
Four pieces of work, each one aimed at a different part of how a model decides who to name.
- 01 ChatGPT Visibility
- 02 Perplexity Visibility
- 03 Google AI Mode
- 04 Entity and Brand Signals
Five models, five ways of choosing a source
They are not interchangeable. One reads the live web, one leans on its index, one leans on what it was trained on, and the work differs accordingly.
-
-
Google AI Mode
Built on the Google index you already compete in
Conventional SEO strength carries over here more than anywhere else
-
Perplexity
Cites sources inline, and shows its working
The most measurable of the five, and the quickest to reflect new pages
-
Gemini
Google data, with its own view of entities
Business Profile accuracy and structured data do real work here
-
Copilot
Microsoft index, inside the tools people work in
Bing visibility and schema clarity are the lever, and both are neglected
The order the work actually happens in
Four stages. Nothing in stage three is worth buying until stage one has been done, and we will say so rather than invoice for it.
- 01
Audit how you are described now
We ask the models the questions your buyers ask and record what comes back: whether you are named, who is named instead, and whether what is said about you is even correct.
- 02
Fix the entity
One consistent account of who you are, what you do and where, across your site, your Business Profile, your schema and every profile that mentions you. Contradictions are what make a model hedge.
- 03
Give it something to cite
Pages that state facts a model can lift and attribute: specifications, prices, coverage, credentials, process. Plus coverage somewhere other than your own site, because self description is weighted lowest.
- 04
Re-test and report
The same question set, monthly, with the answers kept so you can see movement. We show you the misses as plainly as the wins.
Who this moves most
Not everyone needs this, and the ones who need it least are usually the ones being sold it hardest.
-
Considered purchases
Anything people research before they buy. That research increasingly starts as a conversation rather than a search.
-
Firms with a confusing footprint
Trading names, a rebrand, several locations or an old Companies House entry. Models hedge when the record disagrees with itself.
-
Specialists in a crowded market
If you do one thing unusually well, a model can say so, but only if your site says it plainly and somebody else confirms it.
-
Anyone already strong in search
You have the authority. The gap is usually structure and third party description, and that is a cheaper fix than it sounds.
Questions people ask first
Anything not answered here, ask it on the form. You get a person, not a sequence.
Is GEO just SEO with a new name?
No, but it rests on SEO. A model that cannot find or verify you will not name you, so indexation, structure and credibility come first. What is genuinely different is the emphasis on entity consistency and third party description over on-page keyword work.
Can you guarantee ChatGPT will recommend us?
No. Nobody can, and anyone saying otherwise is guessing or lying. The models change, they vary by user and by phrasing, and there is no ranking to buy. What we can do is remove every reason a model would skip or misdescribe you, and measure the result honestly.
How is this measured?
A fixed set of questions your buyers actually ask, run across the platforms that matter to you on a schedule, with the answers recorded. You see which questions name you, which name a competitor, and what changed since last month.
Which platform should we care about?
For most West Midlands businesses, Google AI Mode first, because it rides on the index you already compete in, then ChatGPT on volume, then Perplexity because it is the easiest to verify. We will tell you which after the audit rather than before.
How long does it take?
Entity fixes can show up in weeks, particularly in Perplexity and anything reading the live web. Training weighted behaviour in ChatGPT is slower and less predictable. Treat six months as the honest horizon.
Do we need AEO and GEO both?
Often yes, and they share about half their groundwork, so they are cheaper together than separately. If budget forces a choice, we would start with whichever one your buyers actually use, and the audit tells us that.
Ask the models about your own business and see what they say.
We will run your five most commercially important questions across the major platforms and send you the raw answers, including the ones that name somebody else. No charge and no obligation.
Your email is ready to send
We have opened your email app with the answers already written out. Press send in there and it reaches us.
Nothing opened? Copy the details below into an email to info@nwestradigital.co.uk, or call 07403 641273.