How Did Tally Get 12,000 New Users From AI Mentions in One Week?
It did not, quite. In early 2025, Tally co-founder Marie Martens posted that the form builder had a record week of 12,000 new sign-ups, and that the biggest spike came from AI search. The 12,000 was the week's total, not the number from AI. Over the following months Tally said ChatGPT had become its top referral source. It reported more than 2,000 tracked sign-ups a week from AI tools, and ChatGPT accounting for about a quarter of new sign-ups over a few weeks. Tally's own explanation was not a trick. It pointed to years of genuine presence in public discussions, strong reviews, and a free plan people like recommending. The lesson is about being worth naming, not gaming answers.
What did Tally actually report?
| Date | Source | What Tally said |
|---|---|---|
| Early 2025 | Marie Martens on LinkedIn | A record 12,000 new users signed up in a week; the biggest spike was AI search; still working out what triggered it |
| Spring 2025 | Marie Martens on LinkedIn | AI search models had become Tally's largest acquisition channel; about 25% of new sign-ups in recent weeks came directly from ChatGPT |
| June 2025 | Tally blog, "From 2 to $3M ARR" | ChatGPT became the number one referral source; over 2,000 new users a week signing up via AI tools, "and those are just the ones we could track" |
| January 2026 | Tally blog | Revenue doubled in 2025; ended the year with 12,000 paying customers |
| 2026 | Tally posts | Passed $5M ARR; over 100,000 new users a month |
Note the second 12,000: by the end of 2025 Tally had 12,000 paying customers. The two figures are easy to mix up, and some retellings merge them.
Why did AI assistants start recommending Tally?
Tally's June 2025 blog post offered a theory rather than a proven cause. It linked ChatGPT's rise to its top referral source to a ChatGPT update launched in April 2025, which in Tally's view made it better at surfacing newer or niche tools. The record week itself came just before that update, and Tally said at the time it was still working out the trigger. It also became more likely to suggest alternatives for queries such as "simple or clean form builders", and it browsed the web more often. With browsing, forum threads, blog mentions and other user-generated content became part of what ChatGPT read.
Tally said it had spent years showing up in exactly those places: "sharing what we learn, answering questions, and being human". Its conclusion was that authentic presence in public conversations was paying off. We have not verified Tally's figures independently; this page reports what the company has published.
What made Tally easy to recommend?
Several things Tally has described make it a natural answer to "what's a good free form builder?":
- A generous free plan. Anyone can use Tally for free with unlimited forms and submissions. That gives people a reason to recommend it, and gives assistants a clear, distinctive fact to repeat.
- A built-in referral loop. Free forms carry a small "Made with Tally" badge, which Tally says turns users into advocates.
- Strong reviews. Tally has many reviews on G2 and Product Hunt, where it launched in March 2021. Its AI information page quotes G2 scores against Typeform.
- Years of community participation. Founders and staff answered questions and shared what they learned in public, with a community lead on the team.
- A clear position. Tally describes itself as simpler than Google Forms to customise and cheaper than Typeform, which matches how people ask.
None of these is an AI tactic. They are reasons for people to mention a product, and AI assistants repeat what people say.
What is Tally's AI information page?
Tally publishes a page at tally.so/ai-info that it says is built for AI crawlers and assistants. It lists basic facts, core features, who Tally is for, trust signals from reviews, clients and case studies. It also lists limitations, such as no mobile app and no live team collaboration, plus short guidance on how to describe Tally.
The honest limitations section is the most useful part to copy. It gives assistants accurate boundaries to repeat, which reduces the chance of an answer overselling you. There is no public evidence that the page itself caused Tally's growth; it came after the spike. Treat such a page as a tidy source of facts, not a ranking lever.
What does the measurement lesson show?
Tally's own numbers show how hard AI referrals are to count. It tracked more than 2,000 sign-ups a week from AI tools, but said its onboarding survey, which asks new users how they found Tally, showed "the actual number is a lot higher". Many people read a recommendation in ChatGPT, then type the name into a browser or search for it, and arrive as direct or branded search traffic.
Two practical steps follow:
- Ask new users how they found you, with AI assistants as an explicit option.
- Compare branded search and direct sign-ups with tracked AI referrals over time, rather than relying on referral data alone.
Tools and methods are covered in which tools teams use to track AI referrals and attribution, how to track ChatGPT and other AI referrals in GA4 and how to track leads that came from AI answers.
Can other companies repeat it?
Parts of it, with caution:
- Category matters. "Best free form builder" is a question people ask assistants every day, and there are many tools to choose from. Categories with fewer buying questions will see smaller effects.
- It took years. Tally's community presence and reviews built up from 2020 onward. There was no single campaign that produced the spike.
- Timing helped. Tally benefited when ChatGPT's behaviour changed. The next change could favour someone else.
- It is one company's account. Tally's causes are its own best guess, and the company said it was still working out what triggered the spike.
What transfers is the approach: be useful in public places where your buyers ask questions, earn honest reviews, give people something distinctive to say about you, and publish clear facts about what you do and do not do. How founders can contribute is covered in what a founder should publish to build AI search visibility, and the broader playbook in how to get a B2B software product recommended by AI assistants.
What should you do first?
- Ask ChatGPT and other assistants the questions your buyers ask, and note who is named. See how to check whether ChatGPT recommends your SaaS.
- Find the sources they cite, such as review sites, comparisons and discussions.
- Take part in those discussions honestly, saying who you are, and answer questions even when your product is not the answer.
- Ask every customer for an honest review in their own words, on the platforms that matter in your category and allow requests.
- Publish a clear facts page, including limitations.
- Add an attribution question to your sign-up flow.
What might this look like in practice?
Consider an illustrative scheduling tool, Ruddby, with a free plan and a small team. Its founders spent a year answering scheduling questions on community forums and in a newsletter, always disclosing who they were, and asked every new customer for an honest review. They added "AI assistant" to their sign-up survey. Six months later, the survey showed 18% of new users naming ChatGPT or another assistant, while referral tracking showed only 6%. The team kept its community work and treated the survey figure as its main measure of the channel.
What do people ask about Tally and AI search?
Did Tally get 12,000 users from ChatGPT in one week?
No. 12,000 was the total for a record week. Tally said AI search was the biggest source of the spike.
How many sign-ups does Tally get from AI?
In mid-2025 Tally reported over 2,000 tracked sign-ups a week from AI tools, and said the true number was higher.
What did Tally do to appear in ChatGPT?
It credits years of genuine community participation, reviews and a generous free plan, not a specific AI tactic.
Does an AI info page help?
Tally publishes one, but there is no evidence it caused the growth. It is useful as a clear, honest facts source.
Published by The Kovmere Team · updated Sep 29, 2026 · More resources