If there’s a better use of your budget, I’ll tell you.
Generative engine optimization services
GEOthatearnsyourbrandaplaceinAIanswers
Your buyer asked ChatGPT for a shortlist. Were you on it?
I get your brand named, recommended and cited when buyers ask ChatGPT, Perplexity, Gemini and Claude who to trust. I measure it the honest way, over many runs, and tie it to pipeline.
51%of B2B software buyers now start research in an AI chatbot more often than in GoogleG2, 2026
Under 1 in 100chance an AI tool gives the same list of brands twice for the same questionSparkToro, 2026
Up to 40%more visibility in AI answers from changes to how a page is writtenLab study, KDD 2024
Theshortlistgetswrittenbeforeanyonevisitsyoursite
Buyers used to search, open six tabs and compare. Now plenty of them ask an AI assistant for the three best options and start there. The assistant does the comparing, writes the shortlist and gives reasons. You find out you weren’t on it when the deal never shows up.
The first two rows are the reason to care. The last two are the ones most GEO pitches leave out. The answers move from run to run, and the sources AI leans on can change in a matter of weeks.
One screenshot of ChatGPT naming you proves about as much as one good hair day.
Theupside:AIwillrecommendabranditjustmet
That 33% is the most useful number on this page. A third of those buyers chose a company they had never heard of, because an assistant vouched for it. That’s how a smaller brand gets onto a shortlist next to the category leader.
And what you publish matters. The researchers who named generative engine optimization tested how changes to a page’s writing affect its visibility in AI answers, and found gains of up to 40% in a lab test, not in live ChatGPT. Clear, specific, well-sourced pages get used. Vague ones get skipped.
Every AI answer draws on two things. What the model already remembers, and what it looks up at the moment of the question. They move at very different speeds, and most GEO work only makes sense once you split them.
What the model remembers
What the model looks up
Where it comes from
The text the model was trained on, often months old
A live web search when the question is asked, in ChatGPT search, Perplexity, Gemini and others
How fast it changes
Slowly, as new model versions come out
Within weeks of a page being crawled
What moves it
How often and how consistently the web described you before the cutoff
Pages that answer the question clearly, crawlers that can reach them, and your presence in the sources AI trusts right now
What I work on
Consistent company facts, third-party mentions, reviews and profiles
Page structure, AI crawler access, Bing indexing, citation sources
Most results you’ll see in the first quarter come from the right-hand column. The left-hand column is the long game. Anyone who promises to get you into ChatGPT’s training data by next month is selling a very confident guess.
WhereGEOusuallygoeswrong
Tracking by screenshot. AI rarely gives the same answer twice, so one good result tells you almost nothing. Visibility has to be measured over many runs.
Chasing the model’s memory first. It’s the slowest lever and the hardest to prove, and the faster wins usually sit in what AI looks up.
Writing for robots. Pages stuffed with keywords and “AI-friendly” filler read badly to people, and AI tends to pass them over for pages with real facts.
Ignoring what others say about you. When someone asks for the best option, AI leans on reviews, comparison articles, forums and videos. If you’re absent there, you’re usually absent from the answer.
Stopping at the mention. Being named is nice. Being named for the prompts buyers use right before they buy is what pays.
HowIgetyouontheshortlist
The work runs in five parts. The first starts on day one, the next three run side by side, and the last one repeats every month.
I build a set of the prompts your buyers use when they’re close to a decision. Best options for a use case, you versus a competitor, alternatives, who to hire.
Each prompt runs several times on ChatGPT, Perplexity, Gemini and Claude, so the results show a pattern you can act on.
Repeated runs on every assistant, with mention and citation tracking
Competitor share of recommendations
Citation source analysis
A check on what AI gets wrong or out of date about you
On your desk
How often each assistant recommends you, next to your top three competitors
The sources behind every answer, and the ones you’re missing from
A list of wrong or stale facts AI repeats about you
A 90-day plan ranked by what should make you money first
Weeks 2 to 8
Fix what AI looks up
When an assistant searches the web, it pulls passages from pages it can reach and trust. I make sure AI crawlers can get to your key pages, that those pages work without JavaScript, and that you’re indexed in Bing as well as Google, since some assistants lean on Bing.
Then I rewrite the pages that should be cited so each one answers a real buyer question with specific facts.
On my desk
AI crawler access (OAI-SearchBot, GPTBot, PerplexityBot, ClaudeBot)
JavaScript rendering and Bing indexing checks
Comparison, alternatives and best-for pages
Rewrites with specific facts, figures and sources
On your desk
Key pages every assistant can read in full
Pages built for the prompts buyers use to decide
Passages that still make sense when AI lifts them out
Weeks 2 to 12, then ongoing
Fix what AI remembers
If your site, your LinkedIn and a 2019 directory listing all describe you differently, AI blends them into something nobody at your company would sign off on.
I settle on one clear description of who you are, what you sell and who it’s for, then carry it through your About and author pages, your schema and the profiles AI already reads.
On my desk
Entity SEO and one consistent company description
Organization schema with links to your official profiles
About and author pages
Profile, listing and review platform cleanup
On your desk
One consistent story about your company across the web
Profiles that say what you do today, not what you did in 2019
Ongoing
Get into the sources AI trusts
For your buying prompts, I map the review sites, roundups, directories, forums and videos AI keeps citing, and rank them by how often they show up. Then we work down the list.
Some you can claim or correct in an afternoon. Others take outreach and patience. I’m not a PR agency, so I’ll tell you which is which and who should own each one.
On my desk
Citation source mapping for your buying prompts
Review platform and directory profiles
Roundup and comparison gap analysis
Reddit and YouTube presence review
On your desk
A ranked list of the sources AI cites in your category, with your gaps marked
Claimed and corrected profiles on the platforms that matter most
A plan for the rest, with an owner for each item
Monthly
Re-run, report, adjust
Every month I re-run the full prompt set, compare it with last month, and move effort to what’s working.
Howyou’llknowit’sworking
AI answers vary, so I measure them the way you’d run a poll. Every month you see:
Your share of recommendations for each prompt group, measured over many runs, against your competitors.
How often assistants cite your pages as a source, and which pages.
AI referral traffic from ChatGPT, Perplexity and others tracked as its own channel in GA4, with the leads it produced.
A “how did you hear about us” field on your forms that names the AI tools, because plenty of buyers read the answer and never click.
Some of this will always be an estimate, because most AI platforms share very little data. I’ll tell you which numbers are solid and which are directional.
Yourreputation,readaloud
AI assistants are summarizing what the web says about you, in a confident voice, to people about to spend money. Your reviews, your listings, a comparison article from two years ago and a forum thread you’ve never seen all go into that summary.
That’s why I treat generative engine optimization as reputation work more than a technical trick. The on-site fixes make you easy to quote. The off-site work gives AI a reason to recommend you. It needs both, and it needs real proof to work with. If there’s nothing true and good to say about a company yet, no amount of optimization will make AI say it.
Resultsworthputtinganamebeside
Different markets, same standard. SEO should make you money, and these did.
Longevity and diagnostics
Jinfiniti Precision Medicine
356 to 9,365 first-page keywords
A founder’s peer-reviewed research, made legible to Google. Monthly organic sessions rose from 5,600 to 78,000.
Chateau Mojo is my SEO and AI search consultancy. I’m Jeremy McWilliams.
I’ve spent 12 years in SEO, including time at an award-winning Montreal agency. My work covers ecommerce, health and wellness, biotech, B2B, and other markets where buyers research hard before they commit.
Most AI SEO companies hand you to an account manager after the sales call. Here, the person on the call is the person doing the work.
Buyers in your category ask AI for recommendations, comparisons or alternatives.
You have real proof, like reviews, customers, results or original data, that AI could repeat.
Your SEO basics are in place, or you’re ready to fix them at the same time.
You can give it a quarter for the first changes and longer for the rest.
It’sapoorfitif
You want a guaranteed spot in ChatGPT.
You need a PR agency.
Your company is so new that nobody has written about it yet.
If you’re that new, reviews and SEO consulting will do more for you first. If generative engine optimization is the wrong spend for you right now, I’ll say so on the call.
Generative engine optimization services help a company get named, recommended and cited when people ask AI assistants like ChatGPT, Perplexity, Gemini and Claude about its category. The work covers your own pages, what the rest of the web says about you, and how you measure the results.
What does a GEO agency do?
A GEO agency measures how often AI recommends you, finds out why it recommends others instead, and fixes the causes. That usually means page rewrites, AI crawler access, consistent company facts, and a presence on the review sites and roundups AI cites. A good one also reports it over many runs, since single answers vary.
What’s the difference between GEO and SEO?
SEO earns rankings in search results, and GEO earns a place inside the answers AI assistants write. They share a lot, because AI often searches the web before it answers. But AI tends to weigh mentions, reviews and clear facts more heavily than links, so the two need separate work and separate tracking. The GEO vs SEO guide goes further.
What’s the difference between GEO and AEO?
GEO is about being recommended and cited in answers that AI assistants write, and AEO is about being the direct answer to a question, mostly in Google’s AI Overviews and AI Mode. They overlap, and most companies need some of both. My AEO services page covers the Google side.
How do I get my brand recommended by ChatGPT?
Make it easy for ChatGPT to find, understand and trust you. That means pages its crawler can read, one consistent description of your company across the web, and mentions on the review sites, comparisons and forums it cites for your category. Then measure over many runs, because one answer tells you very little.
How long does GEO take to work?
Answers that come from a live web search, in ChatGPT search, Perplexity and Gemini, can change within weeks of a page being recrawled. What a model remembers from training changes slowly, as new versions come out. I’d expect the first measurable shifts within a quarter, and I’ll tell you which prompts should move first before we start.
How much do GEO services cost?
Engagements run on a monthly retainer. The AI Visibility Audit is the starting point. You get the scope and the monthly fee in writing before you commit.
Can you guarantee ChatGPT will recommend us?
No. Nobody controls what an AI assistant says, and the same question can get a different answer five minutes later. What I can do is measure where you stand, fix what’s in your control, and show you the trend over time.
How do you measure GEO results?
I run a fixed set of buyer prompts several times on each assistant and record who gets named and which pages get cited. Your share of those answers is the headline number. Then I connect it to AI referral traffic and leads in GA4, and to what buyers say in your “how did you hear about us” field.
Do backlinks still matter for GEO?
Yes, but less than you’d think. Links still help your pages rank, and AI often searches before it answers. But when AI decides which brands to name, mentions on trusted sites and consistent facts about you tend to count for more than raw link numbers.
Do I need an llms.txt file?
It’s optional. I’ll add one if you want it, but I won’t bill you for it as if it were magic. The evidence that it changes what AI cites is thin, and there are better uses of the first month. The llms.txt guide covers who actually reads it.
Should I hire a GEO consultant or a GEO agency?
It depends on who you want doing the work. Large generative engine optimization companies bring bigger teams. A GEO consultant brings one person who knows your account end to end. With me, the person on the sales call is the person running your prompts and rewriting your pages.
Findoutifyou’reontheshortlist
Bring me your site and the three competitors you lose to most. You’ll leave the call knowing how often AI recommends you, who it names instead, and what I’d fix first.
