Skip to content
Pricing
Get started free
For in-house marketing

The number that survives the meeting.

Someone senior will ask whether you show up in AI answers, and whether the number is real. Walk in with where you stand, what moved and what it’s worth, with proof that holds when they push back.

Free forever plan · no card · nothing to install
Cross-examination · mention rate Acme · run #23 · 6 Sep
58% of the answers buyers get when they ask about a CRM
Fifty-eight percent of what? of 576 answers · 32 prompts · every tracked engine · run #23, 6 Sep
Up from? +6 pts vs run #22 · 52% one week ago
Is that real, or noise? cleared the band · 95% ci 54–62 against 48–56
Says who? direct · the engines were asked and their answers read · method public
the meeting asks · the console answers survives
The moment

The line in analytics nobody planned for.

Someone senior asked whether you show up in ChatGPT, or sessions arrived from assistants and nobody could explain them. Both start on the same screen.

AI traffic
28 days90 daysBy month
Acme · GA4 · last 28 days · read 6 Sep
by referrerwhole workspace · not per product
Referrersessionskey eventsreading
chatgpt.com 1,612 131 rising · +38% vs prior 28 days
perplexity.ai 604 42 holding
gemini.google.com 318 22 rising · +21%
copilot.microsoft.com 265 19 holding
google.com · AI Overviews not separable · arrives as search
crawled · 1,204 pages fetched by GPTBot, ClaudeBot, PerplexityBot and OAI-SearchBot in 28 days What the engines could read. The docs and the pricing page are fetched daily; the blog was last crawled on 18 August.
The rehearsal

Every question the meeting will ask, answered before it’s asked.

Not four slides you improvise the night before. Each answer below is the product itself, on example data for Acme, a CRM.

Question 1 · 9:41
“Are we even in the answer when buyers ask?”

Yes, in 58% of them, and here is the margin of error. The same questions are asked every time on every engine you track, and every number comes with its proof. The margin shows how sure the number is; it never pretends to certainty.

Visibility
OverviewBy engineBy promptBy stage
Acme · run #23 · 6 Sep · 32 prompts · 6 engines
Mention rate by runnamed in 334 of 576 · 95% ci 54–62
#18 44%
#19 47%
#20 49%
#21 51%
#22 52%
#23 58%
mention rate · named in 334 of 576 answers · 95% ci 54–62 · n 576 · direct Six points up on run #22, and the bands do not overlap. The move is on the pricing prompts; the rest of the suite held.
Question 2 · 9:47
“Who is taking the answers we want?”

HubSpot, in 8 of 32 prompts, and here is the sentence it wins with. Every competitor sits beside you with the answer that named them, and the gap to the leader is a number, not a feeling.

Competitors
OverviewMap
Acme · run #23 · 5 brands tracked · 576 answers
#Brandmention rateshare of voiceendorsednext step
ChatGPT · “best CRM for a 10-person sales team” · HubSpot picked in 11 of 18 answers “For a ten-person team, HubSpot’s free tier is the usual starting point; you pay when you need automation or more than two pipelines.” gap to leader 23 points · the comparison page is the next step on the queue
Question 3 · 9:53
“Did any of the work we paid for actually do that?”

One of three changes moved the number. Two did not, and you know which. Every change you ship is measured before and after. The ones that didn’t work are listed beside the one that did: the budget conversation you can survive.

Answers
OverviewMissedHistory
Acme · 6 runs · weekly · 6 engines
rundatesuitenamedvs previous
#23 6 Sep 32 prompts · 6 engines · 576 answers 58% +6 · cleared the band
#22 30 Aug 32 prompts · 6 engines · 576 answers 52% +1 · inside the band
#21 23 Aug 32 prompts · 6 engines · 576 answers 51% +2 · inside the band
#20 16 Aug 30 prompts · 6 engines · 540 answers 49% series break · 2 prompts added
experimentsa change recorded between two runs, read against the band
runswhat you changedrecordedchangeverdict
mention rate 52% → 58% · 95% ci 48–56 against 54–62 · n 576 both runs The bands do not overlap, which is what a real change looks like here. The move is on the pricing prompts, where the page is now cited in 9 of 18 answers. Run measurementnext run 13 Sep · 32 prompts × 6 engines · priced before it runs
Question 4 · 9:58
“Send me that. I’m forwarding it to the CEO.”

Already written, for a reader who wasn’t in the room. What moved, where you stand and what to do first, with the proof printed under every number. Share a link or save a PDF.

ReportAcme · run #23 · 6 SepExperimentsShareSave as PDF
Acme · AI visibility · September
32 prompts · 6 engines · 576 answers · against run #22
band = 95% ci · n = 576 Named in 334 of 576 answers, and put forward in a third of those. The interval is on every figure because the reader of a PDF cannot hover for it. live · read from run #23 · never recomputed after export
10:04. The meeting ends.

Wednesday, the work is already queued.

The fixes arrive as one list, most important first, sized for an in-house team. The next measurement says whether each one moved anything.

Actions
QueueBlockersSheet
Acme · run #23 · 9 open · 3 closed this month
ActionLeverageDemand (est.)
cited 38 times · names 9 CRMs, none for teams under twenty · the editor is named on the page “TechRadar’s best-CRM list is a good starting point for a small team.” repeated 23 times · the list is the shared source in 3 engines
For the skeptic on the analytics team

The questions after the meeting.

If it isn’t here, the methodology is public, limitations included →

No. Nobody controls these engines, and anyone promising a ranking is selling something they cannot deliver. What is promised is honest measurement: where you stand, what tends to move it, and whether your work did, with the noise separated from the signal.

Walk in with a number you can defend.

A free report on where your brand stands across every engine, intervals included. No card, results in minutes.

Free forever plan · no card · or read a sample report