Case study · The GTM Co-Founder · 23 October 2026

We ran it on our own project first. Here is every number.

The GTM Co-Founder is our founder’s open-source set of go-to-market skills for developer-tool founders, and was #1 Product of the Day on Product Hunt.[1] It had real users and real expertise behind it, and AI answers didn’t know it existed. Here is where it started, what changed in four days, and what we don’t know yet, including the numbers that are still zero.

0of 15 buyer questions where Claude cited it, 20 Oct

66→81AEO audit score, 20 to 23 Oct

30→83citability score, after its first guide went live

14expert answers from its founder, built into the pages

The starting line

Asked 15 buyer questions, Claude cited it in none.

On 20 October 2026, Citeplane asked Claude, with live web search, 15 questions a founder asks before choosing go-to-market help: how to go to market with a developer tool, how to price one, how to get the first 50 users, how to launch on Hacker News. The GTM Co-Founder was cited in 0 answers and mentioned in 0.[2]

Who got cited instead.business.daily.dev appeared in the sources for 7 of the 15 questions, dev.to and GitHub for 5 each, Hackmamba for 4, and Stateshift and Draft.dev for 3 each.[2]
Why.The site was a homepage, a skills list and a quickstart. It had nothing shaped like an answer: no guide, no how-to, no comparison. AI can only quote a page that answers the question.
What the audit said.An AEO score of 66 out of 100. Access, structured data and titles were mostly fine. Citability, how quotable the content is, scored 30.[3]

Step one: the site

Fix what engines trip on, before writing anything.

The plan put site fixes first: page descriptions and structured data, checked in the HTML crawlers actually receive, not only in the browser. They were made in Lovable, where the site is built. The audit moved from 66 to 68.[3] Small, because the real gap was content.

Step two: the expertise

Fourteen answers only its founder could give.

For the three highest-priority questions, Citeplane drafted five interview questions each, about the subject rather than the product: what goes wrong first, which pricing metric gets reversed, where the first 50 users really came from. Fourteen answers were drafted from the founder’s own frameworks and client work, then reviewed and approved by the founder. Anything without a real number behind it was cut, not guessed.

Those answers became the pages’ first-hand material. The pages AI cites today in this category summarise each other. An advisor’s decision rule, like testing whether one developer can get value alone in under an hour, and who signs the cheque, is the kind of specific those pages don’t have.

Step three: the checks

What the fact-check caught before anyone published.

Credit claimed beyond the source.The drafts credited the founder by name for sentences the writer had added. The check flagged it: the answers established the founder as the source of the advice, not of every example on the page. The credit was narrowed to what the founder said, and Citeplane now records who answered each interview.
A fact placed where it implied more.A pricing draft put a vendor’s five usage meters right after advice to buy a pricing study for complex billing, implying that vendor needed one. The source said nothing of the kind. The sentence was cut.
A ranking nobody made.A first-users draft called templated outreach “the other common failure”. The founder had named one most common mistake, not two. The ranking came out.

None of these were typos. Each was a sentence that read well and would have survived a skim, and an engineer who knew the subject would have caught it.

Step four: published

The first guide went live on 23 October.

How to go to market with a developer tool was approved by the founder and published with its structured data: the article, its breadcrumbs and an FAQ block. It went into the sitemap and llms.txt, the text is in the HTML crawlers receive, and indexing was requested in Google Search Console.

The same day, the audit score went from 68 to 81, and citability from 30 to 83: the site now has a page shaped like an answer.[3]

What we don’t know yet

Whether AI cites it. We’ll show you either way.

A better audit score is not a citation. Answer engines have to find the page, crawl it and choose it, and that can take weeks. Citeplane asks the same 15 questions every week, so the next runs will show whether the guide starts appearing in the answers, and who it displaces. We’ll update this page with the result, whatever it is.

First run after publishing, 23 October: still cited in 0 of the 15 answers, but mentioned in 1, for “AI agent that helps founders with go-to-market”.[2] Claude found it through a third-party tools directory, not the new guide. That is worth knowing: the pages engines draw on include the places that list you, not only your own site.

Next: guides on pricing a developer tool and getting the first 50 users are drafted and checked, and four more pages are in interview.

See your own starting line.

The free AEO audit scores how well AI can read and quote your pages, then shows the fixes in order. It takes about a minute, with no email.

Run my free audit →

[1]The GTM Co-Founder, ranked #1 of the day on Product Hunt on 8 August 2026: producthunt.com/products/the-gtm-co-founder. Open source, MIT licensed: github.com/AIDevGTM/gtm-cofounder

[2]Citeplane tracking runs for gtmcofounder.com, 15 buyer questions asked to Claude with live web search: 20 October 2026, cited 0 times and mentioned 0 times; 23 October 2026, cited 0 times and mentioned once. Source counts are the number of questions whose answer cited that domain.

[3]Citeplane audits of gtmcofounder.com: 66 on 20 October, 65 on 21 October, 68 on 22 October and 81 on 23 October 2026. Citability scored 30 on 22 October and 83 on 23 October.