Free Plan vs Paywall: Why Tally Gave Away 99% of Its Product
Short answer: Use a generous free plan when free users can reach value cheaply, naturally expose the product to new buyers and have a clear paid reason to upgrade. Put the paywall around capacity, control or commercial requirements, not the first useful result. Keep freemium only if retained paid contribution covers free-user infrastructure and support within a limit you set in advance.
A paywall can create revenue earlier, but it can also interrupt the mechanism through which a collaborative or shareable product spreads. A free plan can widen that loop, but a large user count can hide weak economics.
The decision is not whether free feels generous. It is whether free usage creates qualified distribution at an affordable cost.
Tally's unusually broad free plan
Tally launched its minimum viable product in September 2020. Founders Marie Martens and Filip Minev manually contacted hundreds of creators and founders. Before a March 2021 Product Hunt launch, the form builder had about 1,500 users and US$380 MRR; the launch roughly doubled its user base.
In Tally's first-year account, the founders reported 11,000 users, 200 paying customers and US$5,000 MRR by October 2021, with 99% of features free and no paid marketing. A later growth account reported 36,560 users, 775 paid accounts and US$20,700 MRR by June 2022.
Those snapshots do not reveal free-user infrastructure cost, cohort conversion, abuse or the exact acquisition share from forms carrying Tally's brand. They do establish that a broad free tier and paid business coexisted.
My view is that most early software businesses paywall too soon. They protect features before proving a distribution loop. The exception is a product with high marginal cost, regulated exposure or no natural upgrade event. There, free can consume the runway without creating demand.
Use the Free-to-Paid Capacity Ratio
The Free-to-Paid Capacity Ratio asks how many retained free accounts one retained paid account can economically support.
Calculate it monthly:
Paid contribution per account divided by free service cost per active account = supportable free accounts per paid account.
| Input | Include | Exclude | |---|---|---| | Paid contribution | Subscription less payment, hosting and paid support | Total revenue before direct cost | | Free service cost | Usage-based infrastructure, attributable support and abuse | Sunk development cost | | Active free accounts | Accounts completing the core job in the month | Dormant registrations | | Paid accounts | Paying accounts still active after refunds | Trials and overdue accounts |
If a paid account contributes £22 and an active free account costs £0.11, the ratio is 200 to 1. That does not mean you should use all capacity. Reserve room for cost spikes, support and taxes.
Identify the product's spread event
Free works best when completing the job exposes the product to someone else. A public form, booking page, report or shared file can carry attribution to qualified users. An internal accounting utility may have no comparable loop.
Map the chain: active free account, outputs shared, recipients reached, recipient visits, new active accounts and eventual paid accounts. If you cannot observe or reasonably attribute that chain, treat free as a product experience choice, not an acquisition engine.
Do not force promotional marks into contexts where they damage the customer's trust or breach contractual requirements. Offer paid removal where brand control has real business value.
Put payment around expansion, not frustration
A free user should complete one meaningful job. Paid value can then attach to higher volume, collaboration, custom branding, administration, advanced controls, service or commercial risk reduction.
Avoid arbitrary restrictions that make the free product appear broken. They may increase short-term upgrades while lowering recommendation and retention. The pricing guide at [how to price a new product or service](/business-finance/how-to-price-a-new-product-or-service/) explains why the metric should rise with customer value.
Ask ten paying customers what changed at the point of upgrade. Their answers should cluster. If one paid for volume, another for a tiny feature and eight upgraded merely to support you, the boundary is not yet dependable.
Measure cohorts, not the headline user count
For each monthly cohort, record free activation, 30-day active retention, upgrade, paid retention, support minutes and infrastructure cost. A low upgrade percentage can support a good business if acquisition is inexpensive and paid contribution is high. A high upgrade percentage can still fail if free usage is costly or the paid tier churns.
Use active accounts as the denominator. Counting people who registered once depresses conversion and obscures unit cost. Also distinguish upgrades caused by usage from those caused by a launch promotion.
Data, consumer, tax and recurring-payment rules differ across jurisdictions. State prices, renewal, cancellation and data use clearly, and obtain qualified local advice for your particular service.
Worked example: FormFolio
FormFolio provides forms for community groups and small companies. These figures are illustrative.
It has 8,000 active free accounts costing £0.07 each in infrastructure and £0.04 each in attributable support. Free service cost is 8,000 x £0.11 = £880 monthly.
There are 160 paid accounts at £29 monthly. Paid revenue is 160 x £29 = £4,640. Payment, infrastructure and paid support cost £7 per paid account, so paid contribution before the free tier is 160 x (£29 minus £7) = £3,520.
The ratio is £22 divided by £0.11 = 200 supportable free accounts per paid account. Actual free-to-paid active accounts are 8,000 divided by 160 = 50. Paid contribution covers the free service cost four times.
| Monthly result | Amount | |---|---:| | Paid contribution before free cost | £3,520 | | Free service cost | £880 | | Contribution after free cost | £2,640 |
FormFolio finds that 35% of new activated accounts originate from links on published forms. It tests moving response exports behind the paywall. Upgrades rise by 14, but free activation falls 22% and the test produces 95 fewer activated accounts from shared-form links. Historical cohorts show that 20% of those activated accounts become retained paid accounts, so the expected lost contribution is 95 x 20% x £22 = £418. The extra 14 upgrades add 14 x £22 = £308, which does not replace the damaged acquisition loop.
It restores exports and instead charges for custom domains, shared workspaces and removal of FormFolio attribution. The test shows why conversion alone is the wrong optimisation target.
Decide when to tighten or remove free
Tighten the plan when free cost grows faster than qualified acquisition, abuse imposes operational risk or paid customers cannot identify a durable upgrade reason. Start with the costliest behaviour, not the most popular feature.
Remove free entirely when the product has no spread event, buyers can evaluate through a guided trial and marginal service cost is material. A time-limited trial may fit better. Compare the options using paid contribution after acquisition and service costs, not registration volume.
Run a six-week pricing test
In week one, calculate active free cost and paid contribution. During week two, interview ten paying and ten retained free users about their completed job and upgrade trigger. Map the spread event in week three.
For weeks four and five, change one boundary for a defined share of new users. Track activation, referrals, upgrades, paid retention and support. In week six, keep the version that increases contribution without destroying the acquisition loop. If neither works, reconsider the upgrade metric before adding more restrictions.
Related guides
Frequently asked questions
What percentage of free users should become paying customers?
There is no universal target. Conversion depends on the market, definition of active, price, free boundary and acquisition source. Build the required rate from your economics. Divide monthly free-tier cost plus desired contribution by contribution per paid account, then compare the resulting paid-account requirement with active free users. A 2% conversion rate can work when free cost is tiny and paid retention is strong. A 10% rate can fail when support and acquisition are expensive. Label outside benchmarks as indicative and check them against your own cohorts.
Should I offer a free plan or a free trial?
Choose a free plan when users can receive continuing value cheaply and their use helps distribution or later expansion. Choose a trial when marginal cost is meaningful, the product has no natural sharing loop or the buyer can evaluate it within a normal decision period. A trial creates a deadline but may not suit products that need data or team adoption over months. Test both on comparable cohorts. Measure activated paid customers and 90-day contribution, not trial starts. Make expiry, charges and cancellation clear under the rules that apply to your customers.
Can I change what existing free users receive?
Usually, but the commercial and legal answer depends on your terms, promises and jurisdiction. Give reasonable notice, explain the change and offer a practical route to export data or choose a plan. Grandfathering can protect customer trust but creates additional product and support complexity. Model the cost of retaining old entitlements against expected upgrades and churn. Do not manufacture urgency with misleading deadlines. Consumer, contract and data-portability obligations vary, so obtain qualified local advice before materially reducing an established service.
How do I calculate support cost for free users?
Tag support contacts by plan and multiply handling time by the fully loaded hourly cost of the people responding. Add tools that vary with contact volume. Then divide by monthly active free accounts, not total registrations. Include abuse investigation and recovery work if it is attributable. Do not allocate all general product development to free users, because much of it benefits paid customers too. Review the cost by behaviour: one high-usage feature may create most of the burden and can be limited without weakening the core free outcome.
What if most paying customers upgrade only to remove branding?
That can be a durable value metric when customers publish work externally and brand control matters. Confirm it through interviews and retention, then price the benefit according to commercial use rather than treating it as an embarrassment tax. Keep the free attribution proportionate and clear. If removing branding is the only paid reason, the business is exposed to a design change or competitor offering. Develop additional paid value around collaboration, administration, capacity or service, but do not stuff the paid tier with unrelated features merely to justify it.
Does a generous free plan make the product look cheap?
Not necessarily. Perception follows the quality of the outcome, clarity of the offer and relevance to the buyer. A broad free tier can signal confidence and reduce evaluation risk. It can also confuse business buyers if paid governance, service and reliability are not explicit. Present the paid plan as a different operating requirement, not simply more buttons. Show exactly who needs it and why. If larger buyers equate free with unsupported, publish service boundaries and response arrangements that you can actually meet rather than relying on prestige pricing.
Comments (0)