Achieving Mission & Profit with Freemium — Erin Webster-Shaller and Paul Apollo, Lose It!

Starting in 2018, after noticing health apps' onboardings getting longer with questions that could not be used for setup, Lose It! A/B tested adding more questions to its onboarding beyond the age/height/weight/goal needed to set a calorie budget. The answers were not necessarily used at first. They kept making onboarding longer until returns diminished. Onboarding is never touched without an A/B test. They also tracked free-user retention to make sure the free experience was not hurt. The rationale: most users have failed at weight loss before, and the longer flow gets them into a committed mindset.

Onboarding lets users set up premium features, first the 'Weekender' calorie-cycling feature ('Do you want more calories on the weekend?'). After the paywall, if the user taps X, a later screen lists the premium features they configured (e.g. intermittent fasting, Weekender) and asks whether to continue on basic without them or start the 7-day free trial. Each feature added to onboarding is A/B tested; only features with immediate day-one resonance are kept.

Tested adding various premium feature setup steps to onboarding, e.g. 'Set a carb goal'. Many did not raise conversion and made the flow more painful, because a carb chart is not something most people care about on day one.

Years ago ran a test locking all macronutrient functions in the app behind premium. Since then the team rarely tests locking features and prefers testing messaging and positioning; new features are assigned to free or premium using a framework (free = enough to lose weight via calorie counting).

Framework for new features: the free product must be enough to lose weight via calorie tracking (calories in, calories out). Macronutrient tracking, advanced diets (keto, intermittent fasting), calorie cycling go to premium, which now has 40+ features. Premium is positioned as what you need if you care about carbs; 'apps and devices' syncing is a big crowd-pleaser. Messaging and positioning are A/B tested rather than locking features.

A/B tests of upgrade messaging compared feature-led copy with copy about how much money the user saves and the value for a low premium price. For a price-sensitive audience the discount messaging wins, so the team leans on it.

2019's big strategic initiative was to grow beyond organic users with a big paid acquisition budget targeting cold audiences. ROAS was projected with curves fitted to each cohort's first 10 days, which predict eventual LTV well for their user base. Many users came in but stayed free.

After abandoning paid, Lose It! priced premium through extensive tests on its organic user base rather than to make Facebook ads pay back. Result is a price roughly half of MyFitnessPal's sticker price.

Marketing focuses on getting people into the free experience, then hands off to the growth team to monetize. Main levers are App Store and Play Store optimization plus organic and word-of-mouth channels; paid is a very small part of the mix.

Automated emails go out at several weight-loss milestones, starting at 5 pounds, asking users to tell their story through a short series of questions and optionally send photos. The stories are then used heavily in social posts, emails and in-app messages. The logic: people in weight loss only talk about it once they see success.

Referral program during a push for premium subscriptions: the referrer received a $10 gift card when the referred person upgraded to premium, effectively a $10 cost per premium subscription. Run on a third-party system with little internal support for experimentation. Treatment vs control compared on LTV.

Research showed base users who have used the app for 30 days without converting are very unlikely to convert (single-digit %). These users get progressively steeper discounts, down from about $40 to as little as $5, framed as an exclusive reward for sticking with the program.

Growth and data science jointly look for leverage points that predict conversion, from spreadsheet analyses to ML models. A model flagged users on a 30-day streak; growth designed a 'streak sale' offering a very discounted premium to that cohort and A/B tested it.

Instead of an email blast, sent the same offer as a shorter in-app message. Test design: take an entire eligible user segment (whether or not they opened the app), split 50/50 between email and in-app message, and measure revenue over the whole eligible cohort. In-app messages also let marketing iterate on copy and inject sales without waiting on engineering to build contextual promos.

Experiments compared delivering in-app messages during the food-logging flow versus after logging is finished or after the user marks the day complete. Post-completion messages celebrate what the user just accomplished and offer something in return.

Users are split into new, reactivated (back after a long absence) and base. Growth spends ~80% of its time on new and reactivated users, hitting them with in-app messages carrying sales while they are still in the app, timed against the known attrition curve. Returning users get celebratory 'Welcome back' messages that acknowledge why they came back, plus first-time/new-user sales.

Sales in in-app messages carry a timer (e.g. 24 hours to take the sale). Because users reflexively X out pop-ups, a persistent countdown timer is shown at the top of the food log for the rest of the day, giving an easy way back to the offer.

Dabbled with in-app network ads in the late 2010s alongside premium subscriptions.

For a number of years company bonuses were paid only if Lose It! hit both its revenue goals and its users' weight-loss goals, keeping monetization aligned with the mission.