Tally is a form builder that gives most of its product away. Its founders say it has 2.5 million users and a team of ten, and its help pages say nearly all features are free. In its 22 September 2026 update the company said 43% of new users now discover it through AI tools.
When almost everything is free, the useful question is not where to put the limit. It is what the few paid things have in common, and what happens to a customer at the moment they decide to pay.
This note reads Tally's public pages, public reviews and the founders' own updates to answer that. We have no access to Tally's data, we have not spoken to anyone there, and we do not know how its conversion is performing. What follows is an outside reading, labelled as one, with five checks you can run on your own product.
What Tally charges for
| What it includes | |
|---|---|
| Free | Unlimited forms and submissions under fair use, payments, signatures, file uploads (10 MB per file), conditional logic, calculations, and integrations with Google Sheets, Notion, Airtable, Zapier, Make and webhooks |
| Pro ($29 a month, or $290 billed yearly) | Branding removal, custom CSS, custom domains and link previews; shared workspaces with unlimited team members; drop-off analytics, Google Analytics and Meta Pixel; unlimited uploads and version history |
| Business ($89 a month, or $890 billed yearly) | Data-retention control, email verification and longer version history |
The pricing page headline shows $24 and $74, which are the yearly totals divided by twelve. The monthly prices are $29 and $89.
Two things stand out. First, sending responses to other tools is free, so access to your own data is not the lever. Second, the paid features sort into four groups: presentation (branding, CSS, domain, link previews), shared access (workspaces and teammates), measurement (drop-off analytics, tracking pixels) and capacity (uploads, version history).
A tempting reading is that people pay once someone else depends on what a form produces. The pricing page does not support that, because the dependency can be served for free. A better hypothesis fits the list: a free form becomes a paid one when it starts representing its owner to other people. Presentation faces respondents, shared access faces teammates, and measurement answers to whoever asked for the results.
The reviews we found point the same way, though they are small and self-selected. Trustpilot reviews ↗ Capterra reviews ↗ One Trustpilot reviewer wrote that they took Pro to host on their own domain. A Capterra reviewer in operations said they upgraded so several team members could build forms. Several Capterra reviewers said the free version limits how far they can customise the look.
We treat this as a hypothesis, not a finding. No public source says why most paying customers upgraded.
Three frictions visible from outside
1. Buying without a preview
Tally's billing page says it does not offer free trials, and its refund page points to the free plan as the way to try the builder. The refund page also excludes "testing of paid features" from refunds, so the intended route is to test on Free and pay with confidence.
Three Trustpilot reviewers in 2024 and 2025 complained about exactly this: paying to try a feature and then being unable to get the money back. Complaint reviews are what unhappy customers write, the numbers are tiny, and no trial and no refund is a deliberate choice that many software companies make. We cannot tell whether it costs Tally any sales.
What we have not checked is what a free user actually sees in the app when they reach for branding removal or a custom domain. If the free plan lets them see the result before paying, this friction mostly disappears. If it does not, it is the clearest public gap at the point of purchase.
2. The organisation boundary
Tally's help pages say an account can belong to only one paid organisation. Joining one moves your existing forms into a private workspace there, and they cannot move back. Leaving removes your access to everything in it, and team members lose access if the organisation's owner does not have Pro. An account switcher was listed on the public roadmap as not started when we looked, with no dates or votes shown. Tally's fair-use policy also restricts sharing outside the registered organisation, so an agency should read the current terms before assuming one subscription can serve several client businesses.
One Capterra reviewer asked for easier workspace management and more admin permissions, and Tally replied that it would share improvements. That is the only direct request we found. We do not know how many customers this affects, so it is a question for users before anyone calls it a bottleneck.
3. The customisation ceiling
The free plan stops where branding and design control begin. That is the product working as designed, and the free plan's generosity is part of the strategy: the founders said in January 2026 that they are optimising for quality rather than revenue, and that they reject enterprise sales calls and demos. Any change that pushed upgrades or gated existing free features would have to overcome that choice.
How the competition sets the line differently
| Product | Free plan, as listed on 27 October 2026 |
|---|---|
| Fillout | 1,000 responses a month, unlimited seats |
| Youform | Unlimited forms and responses under fair use, one user; Pro at $29 a month adds branding removal, custom domains and up to three team members |
| Typeform | 10 responses a month across all forms, with a note that plans can change |
Free plans draw the line in different places. Typeform caps responses hard, Fillout caps them at 1,000 but gives away seats, Youform leaves responses open but limits users, and Tally leaves volume open and keeps shared workspaces for Pro. Because Fillout offers unlimited seats for free, "teams need collaboration, so they pay" is a weak rule on its own. This is competitive context, not evidence about anyone's win rate.
Five checks before you change a limit
Each check says what you would learn and what would show it was unnecessary.
1. Sort your paid features into groups. List everything behind your paywall and ask what the items share. If you find four unrelated groups, your pricing is a feature list. If you find one, you have a trigger you can design around. Wrong if: your customers cannot describe why they paid in terms that match the groups.
2. Buy your own product as a stranger. Walk from the free plan to a paid one without help. Note what you can see before you pay, what you cannot, and what happens if you change your mind. Wrong if: the free plan already shows the paid result and your refund requests are close to zero.
3. Talk to five recent payers and five who left. Ask what they were trying to do in the minute before they paid or gave up. Interviews find mechanisms, not percentages, so do not turn the answers into a rate. Wrong if: the answers scatter with no pattern after ten conversations.
4. Measure sources by what users do next. Tally says 43% of new users found it through AI tools. That figure is self-reported, covers several tools together, is not a share of revenue, and the public material does not define it. A discovery share says where the door is, not what happens inside. Track each source through a published form, a first real external response and a return visit, and look for the sources whose users keep coming back. Wrong if: every source activates and retains about equally.
5. Stress-test the exposure with your own numbers. As an illustration only, if AI sends 43 of every 100 new users and that falls by a quarter, total new users fall by 10.75% with everything else equal. That is arithmetic, not a forecast, and turning it into revenue needs activation and retention data per source. Tally says model changes could alter AI recommendations, and it already invests in AI search visibility, a free MCP server, a referral programme and workflow guides, so none of this is a call to start doing those. Wrong if: your real source mix shows no single source above a level you would worry about.
What we do not know
- Whether people who arrive through AI tools activate, retain or pay differently from other users.
- Why most paying customers upgraded, and how many remove the branding.
- What the 43% figure counts. The founders also cited about 1% of badge-loop signups eventually upgrading in September 2026 and 2% in June 2025. The two disclosures do not establish matching populations, periods or methods, so we draw no trend and calculate no share of the 2.5 million users who pay.
- What a free user sees in the app when they ask for a paid feature.
- Which of these checks Tally already runs.
Our sources are public pages and reviews, read on 27 October 2026. Prices, plans, policies and the roadmap change, so check the live pages before relying on any detail here.
Next step
Product Garage works through live product and go-to-market decisions with experienced operators. This note takes one of those decisions in public, while the thinking is still provisional.
Subscribe to Garage Dispatch to get the next working note.
FAQ
How do you improve free to paid conversion? Start by finding what your paid features have in common, then check what a customer sees at the moment of purchase, and speak to people who paid and people who left. This note describes five checks to run before changing any limit. They are proposals to test, not proven results.
What is a good free to paid conversion rate? We do not offer a benchmark, because the research behind this note contains none that is comparable. Tally's own figures, about 1% of badge-loop signups in September 2026 and 2% in June 2025, come from different disclosures and should not be read as a trend.
How much does Tally cost? Pro is $29 a month or $290 billed yearly. Business is $89 a month or $890 billed yearly. The pricing page headline shows $24 and $74, which are the yearly totals spread over twelve months. Free is free under fair use. Check the live page, because plans change.
Does Tally have a free trial? Tally's billing page says it does not offer free trials, and its refund page points to the free plan as the way to try the builder. The refund page excludes "testing of paid features".
What share of Tally users find it through AI tools? In its 22 September 2026 update, Tally said 43% of new users discovered it through AI tools. The figure is self-reported, covers several tools together and is not a revenue share. The public material does not say whether it counts survey respondents or all new accounts.
Can an agency share one Tally subscription across clients? Tally documents one paid organisation per account and restricts sharing outside the registered organisation under its fair-use policy. That does not rule out every agency use, so read the current terms for your case.