51%
of B2B software buyers start their research with an AI chatbot more often than with Google. Eleven months earlier it was 29%.
G2, The Answer Economy. Survey of 1,076 buyers, March 2026.
When buyers ask AI for a shortlist, are you on it?
Half of B2B software buyers now start their research in an AI chatbot. We run your buyers' questions through ChatGPT, Gemini and Perplexity, count how often each vendor is named, and show you exactly what to fix.
One page, sent by email within one business day. No call, no card, no login to anything of yours.
One buyer question, asked 20 times“What is the best software in our category for a 50-person team?”
Competitor A17 of 20
Competitor B14 of 20
Competitor C9 of 20
Your company0 of 20
The shortlist is now written before anyone visits your site
A buyer types one question and gets three or four vendor names back. If yours is missing, you are out of a deal you never knew existed. Three findings explain why this needs measuring, not guessing.
69%
chose a different vendor than they first planned because of what a chatbot surfaced during research.
G2, same survey.
<1 in 100
is the chance an assistant gives the same list of brands twice. One screenshot proves nothing. How often you appear across many runs is the number that holds still.
SparkToro and Gumshoe, January 2026.
What we measure
Four things decide whether an assistant names you. We measure each one and show the evidence, so you are fixing causes instead of publishing more content and hoping.
- Share of answer
- How often you and each competitor are named across 150 answers: 10 buyer questions, 3 assistants, 5 runs each.
- Who is named instead
- Which vendors take the slots you are missing, on which questions, and how they are described.
- Where the answers come from
- The review sites, comparison articles, forums and pages each assistant cites, and whether you are present on them.
- Whether assistants can read you
- Crawler access for GPTBot, PerplexityBot and Google's AI crawler, and how much of your site still makes sense without JavaScript.
What the numbers look like
An excerpt from the sample report, for an invented company selling time tracking software to agencies. Every figure in a real report links to the raw answers behind it.
| Vendor | Named in | ChatGPT | Gemini | Perplexity | All answers |
|---|---|---|---|---|---|
| Competitor A | 41 of 50 | 38 of 50 | 44 of 50 | 82% | |
| Competitor B | 33 of 50 | 27 of 50 | 35 of 50 | 63% | |
| Competitor C | 19 of 50 | 22 of 50 | 15 of 50 | 37% | |
| Competitor D | 8 of 50 | 11 of 50 | 9 of 50 | 19% | |
| Competitor E | 5 of 50 | 3 of 50 | 6 of 50 | 9% | |
| Example Co (you) | 1 of 50 | 0 of 50 | 2 of 50 | 2% |
How it works
Start to finish on this site and by email. You never have to book a call.
Tell us your category
Your website, what you sell and who you compete with. Two minutes in the form below.
We ask the questions
The questions your buyers type, run five times on each assistant with live web search on.
You get the report
Counts, sources, a crawler check and a ranked list of fixes, with every raw answer attached.
Fix it, or hand it to us
Give the list to your team or agency. Or we do the work and measure again each month.
Pricing
Fixed prices, no contract. Start with the free snapshot. If it shows a gap, the audit tells you why and what to do.
Free snapshot
$0
- 4 buyer questions, 3 assistants, 5 runs each
- You against 3 competitors
- One page, by email, within one business day
- No card and no call
AI Shortlist Audit
$1,500 one time
- 150 answers: 10 buyer questions, 3 assistants, 5 runs each
- You against 5 competitors, question by question
- The sources each assistant cites, and where you are missing
- Crawler access and readability check of your site
- At least 10 ranked fixes, each tied to evidence
- Spreadsheet of every raw answer and a recorded walkthrough
- Delivered in 5 business days
Refunded in full if the report contains fewer than 10 evidenced fixes.
Order the auditMonthly program
$2,000 per month
- We carry out the fixes: pages drafted for your approval, tickets for your developer, listing and review-profile corrections
- The same questions measured again every month
- One written update a month, plain numbers
- Cancel any time
Starts after an audit, so the work is aimed at measured gaps.
Start with an auditWhat we promise, and what nobody can
This field is full of confident claims. Here is exactly where the line sits.
What you get from us
- Every question run at least five times per assistant, with the counts shown.
- Every raw answer handed over, so you can check our reading of it.
- The specific pages and sources behind your competitors' mentions.
- A plain label on each fix: supported by a test, or our best judgment.
- Your money back if the audit has fewer than 10 evidenced fixes.
What we will not tell you
- That we can guarantee a place in any assistant's answer. Answers change on every run, and no one controls them.
- That a position number means anything. We report how often you are named, not a rank.
- That our numbers match the consumer apps exactly. We query each assistant's official API with web search on, and say so in every report.
- That you need a long contract. There is none.
Questions buyers ask us
Is this just SEO with a new name?
The foundations overlap: pages a crawler can read, clear statements of what you do, and other sites that back you up. The difference is what gets measured. In search you hold a position on a page. In an assistant you are either named in the answer or you are not, and the answer changes every time it is asked. We measure that directly and work back to the causes.
Can you guarantee we will show up?
No, and be careful with anyone who says yes. What we can do is find what keeps an assistant from finding, reading and corroborating you, fix it, and measure again with the same questions so you see whether the count moved.
Why do you run each question several times?
Because one answer is noise. Independent research found less than a 1 in 100 chance that an assistant returns the same list of brands twice. How often you appear across repeated runs is far steadier, so that is what we report.
Which assistants do you cover?
ChatGPT, Gemini and Perplexity in every report. Claude and Copilot can be added to an audit on request. We query each one through its official API with live web search turned on, and note that results in the consumer apps can differ.
What do you need from us?
Your website, a sentence on what you sell, and up to five competitors. If you know the questions your buyers ask, send three. We never need a login, analytics access or anything private.
Do I have to get on a call?
No. You request, order and receive everything here and by email, and replies come within one business day. The audit includes a recorded walkthrough you can watch and forward when it suits you.
We already have an SEO agency. Does this replace them?
It does not need to. The audit is written so a third party can carry it out. Many of the fixes are work your agency or content team can do once they know which questions, sources and pages matter.
How does payment work?
The audit is a single card payment through Stripe. The monthly program is billed monthly and you can cancel before any renewal. If an audit contains fewer than 10 evidenced fixes, tell us within 14 days of delivery and we refund it in full.
See your numbers first. It is free.
Tell us what you sell and who you compete with. We run four buyer questions five times on each assistant and send you one page showing how often each of you is named.
Sent within one business day. No call, no card. We use your details only to prepare and send your snapshot.
Prefer email? Write to [email protected].
Who does the work
Lapilo Labs is a small measurement studio run by Cody Rasmussen. You deal with the person who ran your numbers, and every question by email gets an answer within one business day.
Sources. G2, “The Answer Economy: How AI Search Is Rewiring B2B Software Buying,” survey of 1,076 B2B software buyers, March 2026: company.g2.com.
SparkToro and Gumshoe, research on the consistency of AI brand recommendations, January 2026: sparktoro.com.
ChatGPT, Gemini, Perplexity, Claude and Copilot are trademarks of their owners. Lapilo Labs is independent and not affiliated with them.