The AI Visibility Baseline
We ask 50 buying questions about your category to ChatGPT, Claude, Gemini, Perplexity and DeepSeek. Then we send you the report.
How many times they named you. Who they named instead. Why. And our ideas for fixing it, written for your company rather than pulled off a checklist.
$97, back within 2 business days.
30 day money back guarantee. Most companies find they are named in fewer than 5 of the 50.
Why this matters
Your buyers are asking an AI about your category today
It is answering them by name. Somebody in your category is being recommended in those conversations right now, and if it is not you, nothing you own will ever tell you.
4x
Buyers now start here
Citation presence in US ChatGPT prompts went from about 1.6% in June 2025 to about 6.8% in May 2026. Four times bigger in under a year, and still climbing. The question your buyer used to type into Google, they now ask an assistant.
2.5x
And they are better buyers
AI referral traffic converts at about 7.1%, second only to paid search at 7.8% and roughly 2.5 times Google organic. Fewer visitors than search sends you, each one worth considerably more.
0
You cannot see any of it
There is no rank tracker for this. No impressions, no position, no dashboard. Your competitors are being recommended by name in conversations you will never see, and nothing in your analytics will ever tell you. That is what this measures.
2 yrs
The answer is still moving
Models reward sources that are already widely cited, so early positions compound. Once an answer settles, changing it costs what outranking a ten year old page costs today. Right now it is cheap and almost nobody is working on it.
Try it before you buy anything. Open ChatGPT and ask for the best companies in your industry, then count how many times your name comes up. For most businesses the answer is zero, and it has almost nothing to do with their website. The model never visited your site to decide. It read what other sources said about your category.
What actually correlates
The top three signals are all off your website
Ahrefs measured 75,000 brands to find what correlates with showing up in AI answers. Brand mentions beat backlinks roughly three to one, and not one of the top three factors is something you can do on your own site.
Spearman correlation with AI visibility, 75,000 brands. Source: Ahrefs. Correlation is not causation, and we are not going to pretend otherwise. It is still the clearest signal anyone has published about where the leverage sits.
Before you spend another dollar
Three things being sold as AI SEO that do almost nothing
No lift
Schema markup
Controlled before-and-after tests show no causal lift on AI citation. Keep it for Google rich results. Nobody is cited because their JSON-LD is tidy.
0.1%
llms.txt
The share of AI-bot requests that ever touch the file. Google has confirmed AI crawlers do not read it.
Under 20%
Your Google ranking
The overlap between Google’s top ten and the sources AI cites, down from about 70%. Four in five sources an AI cites are not the pages winning on Google.
These are the same findings our free AI Visibility Score already publishes. When we built that scoring model we weighted off-page authority at 55% and schema at 6%, and we argued about whether 6% was too generous.
The method
E-E-A-T got you ranked. It will not get you cited.
Three layers, and the order matters. They are not a metaphor. They are the three strongest measured signals, in order.
Existence
Gets you in the index
Can the model identify you as a thing rather than a string of characters? Without a resolvable record, every mention you earn has nothing to attach to. This is the layer the $97 baseline fixes.
Endorsement
Gets you in the answer
Do sources the model already trusts vouch for you? The strongest measured signal, and the only one that cannot happen on property you control.
Echo
Keeps you there
Does it repeat until it becomes the default answer? One mention is an anecdote. These models are looking for consensus.
Endorsement without Existence leaks, because the model has nothing to attach the mention to. That is why this offer starts at layer one.
What you actually receive
A report about your company, in 2 business days
Nine things, in one document about your company. Not a course, not a dashboard login, not a call.
- 1
Your AI Visibility Score, 0 to 100
Scored separately for each of the 5, because they disagree far more than people expect, plus one blended number.
- 2
The 50 questions we asked, in full
Written for your category, the kind a real buyer types before choosing a company like yours. You see every question, not a summary.
- 3
Every answer we got back, verbatim
All 250 responses, 50 questions across 5 models, with your company name marked where it appears. Usually the striking part is how rarely it does.
- 4
Your count: named in X of 50
One number that tells you whether AI currently sells for you or sells against you.
- 5
The leaderboard: who gets named instead of you
Every competitor the models did name, ranked by how often. This is the page people forward to their boss.
- 6
Why you are missing, specifically
Not "improve your authority". The actual cause for your domain: no resolvable entity record, no third-party mentions, AI crawlers blocked in robots.txt, or a name the models confuse with something else.
- 7
Ideas for fixing it, ranked by impact
Specific, written for your company and your category, not a generic checklist. What to do first, second and third, what each one is realistically worth, and the popular tactics we tell you not to waste money on.
- 8
The Existence module
A short written guide to the one layer you can fix yourself in a weekend: canonical naming, the fact sheet a model can lift verbatim, and disambiguation. Skipping it makes every mention you later earn leak.
- 9
A re-run at 90 days
Same 50 questions, same 5 models, side by side with today. Almost nobody has a starting line, which is why almost nobody can tell whether anything they did worked.
Named in 60 of 245 answers
50 buying questions about our category, asked to 5 models with web search on.
Who got named instead
Why we are missing
Capsule is named in 60 of 245 answers—a respectable showing, but it trails competitors like Pipedrive (162), HubSpot (156), and even Less Annoying CRM (59) by significant margins. The most likely cause is weaker third-party editorial coverage in the sources these models rely on. Capsule appears in sample answers when models list multiple options, but it's rarely positioned as a top pick or given detailed context. This suggests the training data contains fewer authoritative comparisons, buyer guides, and case studies that anchor Capsule as a category leader for small teams.
Read the whole thing: all 50 questions and what the models saidSecure checkout by Stripe. Card, Apple Pay, Google Pay or 50+ cryptocurrencies.
What this normally costs
The tools that measure this start at $95 a month
Published list prices, checked September 2026. The closest comparison is the first row: 50 prompts, the same number we ask, billed monthly.
| Tool | Price | Scope | What you get |
|---|---|---|---|
| Peec AI, Starter | $95per month | 50 prompts | Tracking only. 1 project. |
| Otterly, Standard | $189per month | 100 prompts | Tracking only. |
| Peec AI, Advanced | $495per month | 350 prompts | Tracking only. |
| Otterly, Enterprise | from $1,000per month | custom | Tracking only. |
| Profound | no public priceenterprise | custom | Quote only. |
| The AI Visibility Baseline | $97once | 50 prompts | Plus the diagnosis, the ideas, and the 90 day re-run. |
To be fair to them. Those tools are built for continuous monitoring, which is a real thing and worth paying for once you are actively working on this. We will sell that too. But you should not have to start a subscription just to find out where you stand and what to do about it.
There is also one thing none of them can do. They show you the number and leave you to work out what to do about it. We run a marketplace of 3,800+ vetted outlets, so the report ends with the specific fix and we can execute it for you if you want.
Sources: Otterly pricing, Peec AI pricing. Profound publishes no price publicly, so we do not quote one.
The guarantee
Read your baseline, work through the Existence module, and if it was not worth it, email us within 30 days for a full refund.
What we do not promise. We do not guarantee a position in ChatGPT or any other model. Nobody can. We measure where you stand, and we fix the layer you control.
Who is behind this
We put companies in publications for a living
Essentras runs a marketplace of 3,800+ vetted media outlets with every price in the open. No quote forms, no credit card to browse. That is how we noticed this pattern in the first place: we get to watch what happens after a company appears somewhere.
Questions
How is this different from your free AI Visibility Score tool?
The free tool scores one URL of yours automatically, and it is genuinely useful, so run it first. The baseline is different work: 50 prompts written for your category, run across all 5 models, showing who gets named instead of you. The free tool grades your page. The baseline tells you what the models say when someone asks about your industry.
Is this just SEO with a new name?
No, and the data is unusually clear. Ahrefs measured 75,000 brands: backlinks, the core SEO currency, correlated at 0.218 with AI visibility. Brand mentions correlated at 0.664. The top three factors were all off-site. SEO is still necessary. It is no longer sufficient.
My company is small. Does this only work for big brands?
The research points the other way on one specific point. In the study on citing authoritative sources, the biggest visibility gains went to the lower-ranked sites, with a 115% lift for sites sitting at position five. Incumbents are already in the answer and have nowhere left to move.
Will this be obsolete in six months?
The platforms will keep changing, which is exactly why the method works at the source layer instead of the platform layer. ChatGPT went from about 76% of AI traffic to roughly 53% in a year. Brands that optimised for ChatGPT specifically are rebuilding. Brands that got into the sources show up in Gemini and Claude without touching anything.
Who writes the 50 questions?
Nobody writes them by hand. We read what you actually sell off your homepage, then generate the question set for that specific category, spread across the four stages of a buying decision. Your company name never appears in any of them, because the point is to see whether the models volunteer you. You get every question in the report, and the same set is replayed at 90 days so the comparison is like for like.
What do I actually receive, and when?
The mini-course is available immediately. Your baseline is run by hand for your category and comes back within 2 business days, so that the prompts are ones your buyers would really type rather than something generated from your domain name.
Do I have to buy placements to make this work?
No. The Existence layer is yours to fix in a weekend and the module walks you through it. Placements are how you attack the Endorsement layer, and they are what we do for a living, but the baseline is useful whether or not you ever buy one.
What is the guarantee?
Read your baseline, work through the Existence module, and if it was not worth it, email us within 30 days for a full refund. We do not guarantee a position in ChatGPT or any other model. Nobody can. We measure where you stand, and we fix the layer you control.
Somebody is in that answer today
In front of the buyer who was going to call you. In two years that answer will have settled, and moving it will cost what outranking a ten year old page costs now. Right now it is still moving.