A niche AI app · B2C · host's own app
For a use-it-a-few-times AI app, sell weekly with a 3-day trial
3x revenue from the same number of downloads
Workedpricingpaywall
What they did
The niche AI app first offered a yearly plan with a 7-day free trial; people started trials but cancelled. Switched to a weekly plan with a 3-day free trial. Steve's rule: weekly for one-time-use apps (most AI apps), yearly for long-term-use apps, avoid monthly.
What happened
Same number of downloads, about 3x more revenue.
- Stage
- live app
“we did a 7-day yearly, that wasn't converting as well, cuz people are activating but canceling, David, and then we did a 3-day on the weekly, and then, dude, same number of downloads, 3x more revenues”
“I found that AI apps tend to be more weekly heavy. If it's an app that is a one-time use type of thing, weekly. If it's an app that has a long-term use, I would say yearly.”
Related topics
Get tactics like this every Monday
New tactics from the week's founder interviews, each linked to where it was said.
More from this episode
- Remove the X from the paywall so people must start a trialInstall-to-trial 13% vs 6% with the X; trial-to-paid barely changed
- Label the paywall button 'Continue' instead of 'Start free trial''Continue' did better on iOS; 'Start free trial' did well on Android. One data point
- When people close the 7-day trial paywall, offer 14 free days instead of a discountClients say 14 days worked much better than 7; no numbers
- Coconote: When someone cancels a free trial on the web, offer more free time16% of would-be cancellers took the longer trial
- Mojo: Keep a generous free tier, then show free users a paywall at launch once a weekAbout half of conversions come after day 0 vs ~15-20% industry-wide
- Duolingo: Cut the free trial from 14 to 7 days to get test results twice as fastDoubled the experiment rate and raised conversions overall
- Duolingo: Write 'Try for $0.00' on the trial button instead of 'Try for free'Worked well for Duolingo; not guaranteed for other apps
- Rise Science: On your web funnel, spend time selling the problem before the fixWeb became where the app-first sleep app finally found success; no numbers
- If few people search the App Store for your app, send everyone to a web funnelOver $10K MRR, all from web traffic
- Fabulous: Set up AI agents that watch competitors' apps and ads and write reportsReplaces what would be a full-time analyst job; no numbers
- Weather Up: When someone leaves a hard paywall, send a push offering 7 free daysFew took it and few of those paid; being replaced
- Weather Up: Let people close the paywall, give them 7 free days, then lock the app
- Weather Up: Ship a big update that gets featured by Apple and the press~$100K in the update year with almost no ads; halved the next year without updates
- On Google Play, turn on the grace period and keep retrying failed paymentsBilling-error cancellations drop about a third (31% of cancellations are billing errors)
- Require payment or a trial up front instead of offering a free tierMedian 10% of downloads pay by day 35 vs 2% freemium; retention no worse