Free trial vs. money-back guarantee: what a kids' subscription test showed
A subscription team for children's learning apps tested two growth ideas: moving onboarding to the web before the app, and swapping a free trial for a paid offer backed by a guarantee. The second test is the more instructive one — it traded a small drop in conversion for better revenue quality.
Why a parent sits next to the AI
Teaching.com, founded in 1997, makes reading and typing products for children. Its operating model is unusual for an AI-era learning app: software supports practice, but a parent is expected to sit with the child and judge pronunciation and correctness in real time. The reasoning is practical — automated feedback was not yet reliable enough for that task, and adult help is associated with much stronger learning. For a subscription business, that design also makes the value proposition legible to the paying adult, not only to the child user, which supports a longer-term commitment.
Web-to-app onboarding: trust travels with the brand
Kids' apps face tighter platform limits on tracking and consent tooling, so growth experiments need privacy-safe measurement. Within that constraint, the team tested web-to-app onboarding — starting on the web, then continuing in the app — against the identical flow inside the app. It reported roughly 50% better conversion to trial and 30% better conversion to paid. A plausible explanation offered was brand familiarity: families who already knew the brand for nearly three decades were more comfortable starting on the web. The lesson for other teams is conditional: web onboarding may work best when trust already exists before the first visit.
The guarantee test: fewer conversions, better buyers
The headline test replaced the web free trial with a 30-day money-back guarantee. Headline conversion fell, which is expected when a free option is removed. But the buyers who remained were higher intent: when non-buyers returned to the ad network, the advertising algorithms could find more people like the committed buyers, improving traffic quality. Those users stayed longer, lifting revenue and lifetime value, and refund requests were described as rare. In other words, the guarantee did not win by converting more visitors — it won by changing who converted.
Decision checklist for your own test
Use this checklist before copying the playbook: 1) Confirm the guarantee window, refund steps, and exclusions in plain language. 2) Measure revenue and lifetime value per visitor, not conversion alone. 3) Track refund rate and support load separately from churn. 4) Check whether your brand is already trusted enough for web onboarding to help. 5) For a children's product, confirm every measurement tool is permitted under platform kids' rules. 6) Decide in advance what drop in conversion you will accept for higher buyer quality.
A focus-first team culture behind the experiments
The team also described a deliberately low-distraction culture: asynchronous updates through Basecamp and Slack with notifications off, instant messages reserved for urgent issues, and no rigid Scrum-style framework. Developers were expected to exercise judgement, including late changes when they served users. Whether or not that suits your organisation, it is a reminder that experimentation speed depends on how much uninterrupted focus the people running tests actually have.
| Offer / flow | What happened to conversion | What happened to value | Why it may have happened |
|---|---|---|---|
| Web free trial | Baseline for comparison | Baseline for comparison | Low-friction entry, mixed buyer intent |
| 30-day money-back guarantee (web) | Slight decrease | Revenue and lifetime value increased | Purchase-only funnel sent non-buyers back to ad networks, improving targeting of high-intent users; few refunds requested |
| Web-to-app onboarding | About 50% better to trial | About 30% better to paid | Established brand trust made web start feel safe |
Figures are as reported by the Teaching.com COO in the captured podcast account; treat as dated, attributed case evidence, not independently verified results.
Frequently asked questions
What changed in the Teaching.com offer test?
It replaced a free trial on the web with a 30-day money-back guarantee. Conversion dipped slightly, but revenue and lifetime value rose because purchase-only traffic was higher intent and stayed longer, while few people requested refunds.
How did web-to-app onboarding perform?
Web onboarding converted about 50% better to trial and about 30% better to paid than the same onboarding inside the app, in the team's reported results. The team linked the gap partly to long-standing brand trust on the web.
Why involve parents if AI is used?
Because current AI could not reliably judge a child's pronunciation or give accurate feedback, the product was designed around a parent sitting alongside the child to provide that feedback, alongside AI support.
What constraint shaped experimentation for a kids' app?
Some mobile tooling could not be used in a kids' app under platform rules, so the team had to find alternative ways to collect the information it needed while respecting privacy and consent requirements.