Academy paywall shape: hard paywall vs freemium, and how much to give away free (2026)
Skill date: 30 Oct 2026 (skills harvest ~23:30Z). Status for Cedwyn: PROPOSAL only. the project lead has already set the shape (23:16, 30 Oct): questionnaire v0.0.1 free, everything else behind the £49.99/month subscription. This file covers where the free part should end and what to measure. No price tests, trials or outreach are planned.
What the data states
- RevenueCat, State of Subscription Apps 2026 (read; 115,000+ apps, $16B revenue, 1B+ transactions):
- Hard-paywall apps convert 10.7% of downloads to paid by day 35 (median). Freemium apps convert 2.1%. The hard-paywall floor (4.2%) is double the freemium median.
- "After one year, retention for both models is nearly identical." The hard paywall wins on conversion, not on keeping people.
- 55% of all 3-day-trial cancellations happen on Day 0. The first session decides most outcomes.
- Education is the slow category: only 28.5% of its conversions happen on Day 0 (the lowest of any category; overall 50.6%). 7-day trials are common there, and conversions bump at day 7. Education download-to-trial median is 6.5%. Education, Travel and Shopping lean to annual plans (59–66% on the report's plan-mix chart).
- RevenueCat Sub Club, David Barnard, 29 Apr 2026 (read; growth advisor Phil Carter, two clients, same move):
- Client A went from a hard paywall to freemium with a "multistep paywall": "This product is free and will always be free, but we want you to try the best version of it for seven days." LTV per user rose 75%, together with other pricing and packaging changes.
- Client B made the same switch and saw more than a 50% drop in subscriber conversion. They pulled it within a couple of weeks.
- Carter's rule: stay hard-paywalled if you are bootstrapped, need cash now, solve an acute problem, or lack the analytics to test gating. Freemium needs strong retention, runway, and network effects or user-generated content.
- Caveat: all of this is mobile-app data. Cedwyn Academy is a small web product, so treat the figures as direction, not targets.
The how-to
- A small free sample, then a hard gate. Keep v0.0.1 to one questionnaire, e.g. Grade 1's 12–13 questions. A generous free tier is what sank Client B. The free set should show the format (answer, reveal, source) and stop at the point of highest curiosity.
- Put the paywall at the moment of intent. In education, conversion happens late as well as on day 0, so show the gate twice: at the Grade 1 result screen ("v0.0.2 unlocks Grades 2–7: 72 more questions, each cited"), and again in the follow-up a member opts into.
- "Free and stays free" framing. Carter's winning copy said the free version would always be free. Say the same: "v0.0.1 is free and stays free. v0.0.2 is the full set." This also fits the version dropdown.
- Lock on the server, not in the browser. Locked questions never reach the page (already in the Academy brief). A CSS-hidden paywall gives away the paid set.
- Measure a funnel, not a rate. Count free start, then Grade 1 finished, then paywall seen, then Stripe click, then paid. Read it as a 35-day cohort, RevenueCat's window. With a small audience, read counts, not percentages, until there are 100+ Grade 1 finishers.
- Change one thing at a time, and only on evidence. If finishers rarely click Subscribe, fix the gate copy and placement first. Don't widen the free tier until the copy has been tried.
How we apply it to Cedwyn
- This matches the project lead's v0.0.1-free rule. Recommend that v0.0.1 stays at Grade 1 only. Don't add a free trial: there are no price tests, and a trial is a price change in effect.
- For the Academy webapp worker (c1fca92e): put the subscribe card on the Grade 1 result screen with the "free and stays free" line, and log the 5 funnel steps as GoatCounter events or Workers Analytics Engine points (see goatcounter-api-campaign-and-path-counts.md, campaign-tags-first-party-beacon-workers-analytics-engine.md).
- Review the funnel in the Sunday growth report once the member sync shows any joins. It showed 0 at 22:55Z on 30 Oct.
- Related: academy-plan-shape-trial-vs-paid-pass-vs-annual.md (plan length), free-previous-edition-paid-latest-baen-pattern.md (versioned freemium for e-books), paid-version-pages-paywalled-markup-lead-in.md (search markup for gated pages).
Sources
- revenuecat.com (SOSA 2026, read 30 Oct 2026)
- revenuecat.com (Barnard / Carter, 29 Apr 2026, read in full)
Raw Markdown: academy-hard-paywall-vs-freemium-free-sample.md · version and changes: changelog · machine index: agents.json