What's inside the deep diagnostic
The deep diagnostic is a report in seven sections. We ask 30–50 of your buyers' questions in six assistants — ChatGPT, Gemini, Claude, Grok, Perplexity, Microsoft Copilot — 10 times each. This sample report shows each section: the question it answers for you and what it looks like.
- 30–50of your buyers' questions
- 10runs of each question
- 6assistants
- $249one-time
There is no demo report here: a real report is written for a real client. The examples come from a real free check of allbirds.com on 8 Oct 2026, with its numbers and quotes as they are. The free check asks each question once and re-runs the 1–3 key questions 5–10 times; under each example we say what the diagnostic shows instead.
1. Summary on one page
The question it answers: Where do we stand in AI answers, and what do we do first?
What it looks like: Your two scores, who the assistants name instead of you, and the three actions to take first.
Example from a free check of allbirds.com · 8 Oct 2026
58 / 100
Mention score: Allbirds named in 35 of 60 answers
92 / 100
Site score: what AI crawlers can reach and read
In the diagnostic: The scores rest on 1,800–3,000 answers instead of 60, each with its margin of error. The three actions are written for your goals, from sections 2–6.
2. The assistants' answers
The question it answers: Which assistants name us, for which questions, and how often?
What it looks like: A line for every question and assistant: named, recommended, your site cited — each as «in N of 10 runs».
Example from a free check of allbirds.com · 8 Oct 2026
| Assistant | Named | Recommended | Site cited |
|---|---|---|---|
| ChatGPT | 7 of 10 | 7 of 10 | 1 of 10 |
| Gemini | 3 of 10 | 2 of 10 | 0 of 10 |
| Claude | 7 of 10 | 7 of 10 | 0 of 10 |
| Grok | 8 of 10 | 8 of 10 | 1 of 10 |
| Perplexity | 4 of 10 | 4 of 10 | 0 of 10 |
| Microsoft Copilot | 6 of 10 | 6 of 10 | 2 of 10 |
Out of 10 questions, one answer to each.
One question where Allbirds was not named
“What are the best ethical minimalist shoes for someone who wants a flexible, low-profile everyday sneaker?”
0 of 6 assistants named Allbirds. 5 of 6 named Vivobarefoot.
“Vivobarefoot is the performance pick at $120–$220.”
In the diagnostic: Each question is asked 10 times, so a line reads «named in N of 10 runs» and shows one of four statuses: not named, mentioned, recommended, recommended first. One answer can be chance; ten runs show what holds.
3. Source map
The question it answers: Where do the assistants take their answers from, and are we there?
What it looks like: The sites the answers cite, how often, and whether you and your competitors are on each page.
Example from a free check of allbirds.com · 8 Oct 2026
The assistants built their answers from 137 sites; 6 of them are competitors' own. The Allbirds site was a source in 4 of 60 answers.
| Site | In answers | Allbirds there | Competitors there |
|---|---|---|---|
| ecothes.com | 27 of 60 | Yes | VEJA, Thousand Fell, 8000Kicks, Baabuk, Giesswein, Nike, HOKA, Rothy's |
| thegoodtrade.com | 10 of 60 | No | VEJA, Thousand Fell, Nike, On |
| cobbledgoods.com | 9 of 60 | Yes | Baabuk, Nike |
| sustainablykindliving.com | 9 of 60 | No | VEJA, Cariuma, Thousand Fell, Baabuk, Vivobarefoot, TOMS |
| wetried.it | 8 of 60 | Yes | Giesswein |
In the diagnostic: We read up to 40 of the most-cited pages, not 10 (here 9 of the 10 could be read), and mark every page where competitors are named and you are not. We check open pages: a page behind a login or a paywall is marked as not read.
4. Your site through the bots' eyes
The question it answers: Can the assistants' bots get into our site and read what we sell?
What it looks like: For each family of AI crawlers: let in by robots.txt, and given the same text as a browser. Then what the raw HTML says without JavaScript: an H1, what you sell and for whom, a price, and whether the page can be indexed.
Example from a free check of allbirds.com · 8 Oct 2026
92 / 100
Site score of allbirds.com
| Check | Points |
|---|---|
| Crawlers of OpenAI, Anthropic, Perplexity, Google and Bing let in, with the full text | 60 of 60 |
| Main text and an H1 in the raw HTML | 10 of 10 |
| The H1 and the first lines say what it is and for whom | 7 of 15 |
| A price in the HTML | 10 of 10 |
| The page can be indexed | 5 of 5 |
The H1 “Wildly Comfortable. Super Natural.” is a slogan: it does not say what the product is.
In the diagnostic: A longer list of bots; which of your pages Bing and Brave have in their index; whether Google Search Console and Bing Webmaster Tools are set up; and look-alike domains that could be taken for yours.
5. Penalty risks
The question it answers: Is anything on our site, or done in our name, likely to look like manipulation to a search engine or an assistant?
What it looks like: A list of checks. Each one is marked found or not found, with the page where we found it. No score.
No example — the free check does not run these checks
- Different text for bots and for people — the page a crawler gets is not the page a browser shows
- Pages hidden by mistake — noindex or canonical tags on pages you want found
- Challenge pages — a bot gets a «checking your browser» page instead of your text
- Hidden text and lines aimed at AI — text people cannot see, or lines that tell an assistant what to say
- «Summarize with AI» buttons — buttons that ask the assistant to remember you or recommend you first
- Your own «best» lists — posts on your blog that rank your own product first
- Template pages in bulk — many near-identical pages made for long search queries
- Paid mentions without a label — placements and mentions bought without a «sponsored» mark
- Reviews and experts no one can trace — testimonials or authors that are not real people
6. Plan: where, what, how, who, price
The question it answers: What exactly gets done in the month, where, by whom, and for how much?
What it looks like: A table with one line for each piece of work.
| Where | What | How | Who | Price |
|---|---|---|---|---|
We write this table for your own goals, so it stays empty here.
7. Method and every answer in CSV
The question it answers: Can we check the numbers ourselves?
What it looks like: How we pick the questions, run them and count — on the methodology page. And a CSV file with every answer, one per row: question, assistant, run, time, named, recommended, cited, competitors, links and the full text.
Example from a free check of allbirds.com · 8 Oct 2026
| Question | Assistant | Run | Time, UTC | Allbirds named | Named instead | Answer |
|---|---|---|---|---|---|---|
| “What are the best ethical minimalist shoes…” | ChatGPT | 1 | 2026-10-08 18:16 | No | Saola, Vivobarefoot, Lems | “If your priority is ethical production/materials + a genuinely flexible, low-profile everyday sneaker, I’d focus on barefoot/minimalist brands…” |
| “What are the best ethical minimalist shoes…” | Perplexity | 1 | 2026-10-08 18:14 | No | Vivobarefoot, Feelgrounds, Mukishoes | “For a flexible, low-profile everyday sneaker with credible ethical materials, these are the strongest options…” |
| “Can you recommend an ethical and sustainable sneaker brand…” | Microsoft Copilot | 1 | 2026-10-08 19:48 | Yes | VEJA, Cariuma, Faguo | “Best pick under $120: Allbirds Wool Runners ($98) — carbon-neutral, comfortable, and widely certified…” |
| “What are the best eco friendly sneakers for everyday walking…” | Microsoft Copilot | 1 | 2026-10-08 19:48 | Yes | Ccilu, Delo Go Green | “Best eco-friendly sneakers under $150 for walking and commuting: If you want sustainable, comfortable shoes within budget, Allbirds Tree Runners…” |
Four rows of 60, the answers cut short here; the CSV holds the full text.
In the diagnostic: The free check asks each question once, so almost every row says run 1; the 1–3 key questions are re-run 5–10 times. In the diagnostic each question has runs 1 to 10 — 1,800–3,000 rows, all of them in the CSV.
How to start
- Check your site free
Ordering opens soon. Questions: contact@namedtop.com.