Is this a real report or a mock-up?

Every excerpt on this page came out of a run against the named business on the date shown, and each one links to the full report it was cut from. Nothing here is illustrative data.

Why do the excerpts ask 30 questions when the free report asks 12?

Those runs were made before the free tier was set at 12 questions, and they really did ask 30, so that is what the pages say. A free report today asks 12 and a paid one asks 30 — the same question set, sampled at two depths, producing the same sections either way.

Why is there no score?

Because a number out of a hundred is not actionable and cannot be checked. A score would compress the one thing worth reading — which question was lost, and to whom — into a figure that goes up for reasons nobody can name.

Do you run these against competitors too?

The competitors in a report are not chosen by us. They are whoever the engines named while answering the buyer questions, counted by how many questions and how many engines cited them. That is why the list is sometimes a directory rather than a rival business.

Can I hand this to a client?

A free report carries our logo, which is the paywall. On the Agency plan every report carries your name and logo instead — set once in Settings, printed on every report you open from then on. A standing client portal is being built; the logo is live, and a private link you can send a client is too. Your own brand colour is deliberately not on the list — see the plan comparison.

How long does a report take?

Takes about two minutes. Enter one client website and it runs while you watch; there is no account, no card and no call.

Sample report

What a report actually says

Six sections. Below is each one — what it is and why it is in there — cut from reports that really ran, against named businesses, on the dates shown.

Anatomy

Six sections, in the order they appear

What each one is, why it is in there, and a cut from a report that ran. Every excerpt is from a different business, and links to the run it came out of.

01The verdict

A sentence, and the two counts behind it

The report opens on how often the business came up at all. Two numbers, kept apart on purpose: questions that never name the business, and questions that name it outright. The first is whether buyers find them; the second is whether the engines know who they are when asked directly. A business can be strong on one and absent from the other.

Magnolia Family DentistryAustin, TX2026-09-12Full report

Magnolia Family Dentistry appeared in 4 of 22 buyer questions.

Tested 30 buyer questions across ChatGPT (API), Perplexity, Gemini. Magnolia Family Dentistry appeared in 4 of the 22 that did not name it.

4 of 22: questions that never name them
18: answered entirely without them
8 of 8: questions that named them outright

02The questions you lose

Named, one engine at a time

Each lost question with the domains each engine cited answering it. Per engine, not averaged: they disagree, and the disagreement is information — a question lost on all three is a content problem, a question lost on one is usually a source problem.

SavvyCal2026-09-16Full report

Buyer question	ChatGPT (API)	Perplexity	Gemini
most recommended scheduling software	zapier.com	efficient.app, forbes.com, workflowautomation.net	—
who is the best scheduling software for small teams	calendly.com, acuityscheduling.com	salesforce.com, worldmetrics.org, peoplemanagingpeople.com	peoplemanagingpeople.com, forbes.com, jotform.com
best scheduling software for sales teams	chilipiper.com, blog.hubspot.com	cleverly.co, youcanbook.me, albato.com	revenuehero.io, zencal.io, bestpage.ai

03Who wins instead

Counted, not chosen

We do not pick the competitors. The list is whoever the engines named, ordered by how many questions and how many engines cited them — which is why a directory often outranks every real rival on it. That is the finding, not a flaw in the method.

Abacus Plumbing, Air Conditioning & ElectricalAustin, TX2026-09-14Full report

rotorooter.com13 questions3 of 3 engines
mrrooter.com8 questions3 of 3 engines
rootermanofaustin.com4 questions3 of 3 engines
homeguide.com9 questions2 of 3 engines
reliantplumbing.com9 questions2 of 3 engines

04Where the answers came from

And which of those you can edit

The cited domains, tagged by what they are. The tag is the point: a directory listing is a page someone can claim and correct this afternoon, and a Reddit thread is not. Sorting the sources this way is what turns a citation list into a list of work.

Magnolia Family DentistryAustin, TX2026-09-12Full report

reddit.comCommunity threadcited on 20 questions
zocdoc.comDirectory you can claimcited on 17 questions
healthgrades.comDirectory you can claimcited on 13 questions
bbb.orgDirectory you can claimcited on 5 questions
toprateddentist.comRoundup / aggregatorcited on 14 questions

05What the site never says

The sentence that is missing, written out

For each lost question, what the site would have to state for an engine to be able to quote it. Uncomfortably literal on purpose — engines quote claims, and a page that implies something is a page that gets skipped for one that says it.

Single Grain2026-09-14Full report

most recommended digital marketing agency
The site would need to explicitly say it is one of the most recommended digital marketing agencies or provide a comparable recommendation claim.
who is the best digital marketing agency for ecommerce
The site would need to say directly that Single Grain is the best digital marketing agency for ecommerce, not just list ecommerce SEO as an expertise area.
best digital marketing agency for SEO
The site would need to directly claim it is the best digital marketing agency for SEO.

06What to do

Each one carrying the questions it answers

The number beside an action is how many lost questions it addresses. It is a count, never a projection and never a multiplier — nobody can promise an engine changes its mind, and a forecast here would be the first invented number in the document.

Emmer & RyeAustin, TX2026-09-16Full report

Editanswers 16 questions
Own your listings on opentable.com, tripadvisor.com
On 16 of the questions you lose, the assistants cite a directory rather than any business's own site. Those listings are the answer buyers are being shown.
Editanswers 13 questions
Answer "best restaurant for families in Austin, TX" on /
The site would need to say it is family-friendly or otherwise explain why it is a good choice for families.
Editanswers 12 questions
Add structured data describing the business
The site states nothing machine-readable about what it is or where it is, so assistants have to infer both from prose.
New pageanswers 3 questions
New FAQ entry answering "how to choose a restaurant"
The site would need guidance or criteria for choosing a restaurant.

What is not in it

No score. A number out of a hundred cannot be checked and cannot be acted on, and it moves for reasons nobody can name. Every competitor in this category leads with one, which is roughly the argument against it.

No weekly tracking. Re-running the same questions on a calendar measures engine drift, not your work. The re-test follows a publish instead, so a before and after is attributable to the change that caused it.

No projections. No traffic estimate, no “you could win N more questions”, no percentage lift. The only numbers are counts of things that happened.

Depth, and why these say 30

The runs quoted above each asked 30 questions, and the pages say so, because that is what those reports did. A free report today asks 12 and a paid one asks 30 — the same instrument sampled at two depths. Every section shown here appears either way; the deeper run simply has more rows in it.

All 8 reports, in full

Each one is a different industry and a different named business. Read whichever is closest to a client you have.

FAQ

Questions

Run one on a client site

The same six sections, on a business you already know well. If it tells you nothing you did not know, you have lost about 2 minutes.

Get your free reportOne client website, no signup. Paid plans will ask the same questions 2.5× deeper.