Monetising a free AI app without selling personalisation

• Scanning and logging. Free gets a limited number of scans, and the Foodhak score explains what it means for your health. Pro adds smart swaps, instant logging to the planner and bookmarks.

• Recipes. Free gets one recipe a day. Pro unlocks unlimited, goal-based recipes and lets people tweak protein, carbs or fibre.

There were three ways to structure the trial. I laid them out side by side for the stakeholders.

• No trial: Shape: Free version, optional upgrade any time; Paywall: Soft: inline banners and blurred cards; Pros: Builds trust and habit without pressure; Cons: Slow monetisation curve
• Pure trial: Shape: 14 days of full Pro, then back to free; Paywall: Hard paywall at the end of 14 days; Pros: Lets people feel the value before paying, captures high-intent users; Cons: Higher churn after the trial ends
• Mixed trial: Shape: Pro features visible, unlocked through engagement; Paywall: Soft paywall, plus contextual nudges at high-intent points; Pros: Unlocks depth with engagement and consistency; Cons: Slightly complex logic to automate the triggers

I based the trial length on data. The RevenueCat *State of Subscription Apps 2025* benchmark covers more than 20,000 apps, from January to December 2024.

The benchmark gave me three things to work with:

• Trial length depends heavily on category. Gaming apps mostly favour short trials, with 96.3% lasting four days or less. Education and Health & Fitness apps lean longer, with over 80% lasting five to nine days or more, which fits a product where progress takes time.
• Longer trials convert better, up to a point. Trials of 17 to 32 days had the highest median trial-to-paid conversion, at 45.7%.
• Health & Fitness apps are the most likely to use a mixed trial strategy, at 56%.

My recommendation: Pure trial, then mixed.

Fourteen days of full Pro access lets people see what a week of data looks like and start building streaks. When the trial expires, they drop into the mixed tier. They keep their habits and get nudged to upgrade at points of high intent. By day 14 they've built data and streaks, so losing Pro feels like losing something that's theirs. That's loss aversion, and it works without a hard sell.

The flows were built so the trial length could move to 10 days later. I pitched 14 days. After the stakeholder meeting with the founder, director and product owner, the decision was 10 days.

During the trial, the person has full Pro access. The flow has checkpoints on days 8 to 9, day 10 and day 11, so the nudges land as the trial nears its end.

When the trial ends, people drop into the mixed tier. They keep their habits, and the paywall appears only at high-intent points.

People who stop opening the app need a different path from people who are active. I mapped the pre-trial and post-trial inactivity flows separately.

Cancelling is where trust is decided. The person should always know what happens next, and they should be able to leave without being trapped.

For the Pro user journey I wrote a review on sticky notes: what the person wants to do, what the payment screen looks like, and what they should see at each step. One note stayed with me. Give Pro information a permanent place in Settings, so people can always find out what they have and how to change it.

People reinstall the app or change device, and their purchase should come back without a fight.

A failed payment is a moment of anxiety. I designed each state separately: the offer, the upgrade, the restore, the failure and the success.

The paywall fires at three moments:

• After onboarding.
• When the app is opened once a week.
• After a certain number of app sessions.

The transparency rules sit alongside these: the trial length, the price, the billing (monthly or yearly) and how to cancel. People should know exactly what they're signing up for.

The strongest trigger is a reward the person has already earned. For a highly engaged user, the app shows what their logging has built, and the weekly insight becomes the moment to upgrade.

The same paywall can be opened from three places, and each version should lead with what that person came for. I used Duolingo's paywall as the reference:

• From the shop or hearts, the focus is unlimited hearts.
• From ads, the focus is no ads.
• From the review tab, the focus is personalised practice.

Each gated feature got its own paywall, so the upsell matches what the person just tried to do.

The trial paywall shows the price and the trial length. The entry point comes after onboarding.

I collected paywall examples from other apps to see how they introduce the offer, and which features they put behind it.

The hardest part was the blurred cards. A locked feature shown as a blurred preview tells people the thing exists, which helps conversion. It also makes the app feel like it's keeping things from them.

My recommendation was to swap constant blocking for behaviour-based upsells. Instead of a blurred card on every screen, show a paywall once someone has earned it:

You've logged consistently for 7 days. Unlock your weekly nutrition trends." + CTA

I had two reasons. A blurred card that's always there feels like being left out. And the upsell lands harder when it arrives after the person has shown they care. I have no test results for the alternative, so this stays a recommendation I haven't been able to prove.

• A hard paywall at day one. It asks people to pay before they know what they're paying for.
• A paywall on every screen. It trains people to ignore the upsell, and it makes the app feel like a shop.
• Gating personalisation. That's the product. Selling it would break the promise.

The Pro report is what the summary grows into. Each section answers a question the person has:

• Alignment score. One number that shows how closely the week matched the person's goal, with a seven-day strip underneath.
• Meals. Average calories, how each macro moved against the week before, and what shaped the score.
• Meals x movement. Calories eaten against steps taken, day by day.
• Steps. The weekly total, a bar for each day, and what shaped the step score.
• Sleep. The average sleep, the sleep duration for each night and what shaped the sleep score.
• Mood mix. The moods logged most often, with a food and mood insight.
• Scans. The number of items scanned and the top-scoring items.
• Next up. A short note on what would lift next week's scores.

Meals, steps and sleep each end with a highlight and one thing to do next week.

The report is also shareable. People share a single number or their mood mix, and each card is a reason to come back.

The score has three tiers, and each tier has its own label:

• Mostly Aligned Week at 60.
• On Track Week for scores from 65 to 79.
• Strong Alignment Week at 80 and above.

• Analysis over review. The report explains why a week went the way it did. A list of totals is a log. An explanation is a reason to pay.
• Available to everyone, with no minimum logs. Anyone can see the full report, even after a light week. Requiring a minimum number of logs would have locked out the people who most need the nudge.
• Visually distinct from the free report. The Pro report uses a different treatment, so people can tell at a glance what they're paying for.
• Custom mood. People can either remove the mood section or add their own word for it, so the report can be shaped without a complicated setup.

The rules were the hard part. A macro breakdown only makes sense against the person's goal. A protein-heavy week is a good week for someone losing weight, and an unusual one for someone gaining. The Balanced, Carb-leaning, Protein-forward and Fat-heavy labels only work when the thresholds change with the goal, so the rules are written per goal.

• Macro persona labels the week against the goal's macro ranges.
• Mood mix shows the top three logged moods, sized by how often they appear. Fewer than three different moods means no mood mix. A tie goes to whichever was logged first.
• Power duo pairs two ingredients from the week: one that supported the primary goal and one for the secondary goal.