Mosaic Group · B2C · app portfolio company
Estimate lifetime value as a range of weighted guesses, then narrow it monthly
Lets new products act on LTV before years of data exist; no numbers
Workedanalytics-attribution
What they did
Early in a product's life, run a range of LTV scenarios (e.g. plus or minus 20% in 5% steps), assign probabilities, pick a best estimate and act on it. Re-check every 1-2 months as renewal data comes in, dropping scenarios (six to four to two) until one precise estimate remains, which takes multiple years. At Mosaic's scale an in-house model could predict the next ~20 renewals from the first one or two, since most churn happens in the first renewals.
What happened
Guest says it lets you make decisions under uncertainty; mature portfolio data made predictions reliable.
In their words
Sub Club by RevenueCat · Why Most Apps Hit a Revenue Ceiling (and How to Plan for It) — Patrick Falzon, The App Shop“run a range of scenarios and say, you know, plus or minus 20% and maybe 5% increments, what are my possible LTVs, assign some probabilities to them, come up with your best estimate, start making decisions and acting on that, and then just go back and check it every month or 2 months”
“if we got, you know, an initial renewal point or an initial two renewal points from a given cohort, we could pretty reliably predict out then what the next 20 would look like from there.”
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
- Mosaic Group: Only buy or build apps that could reach tens of millions without outspending giantsBuilt one of the leading app portfolios; sold to Bending Spoons
- Mosaic Group: Build several $30M apps instead of pushing one app past its ceilingGuest's answer to the $10-30M ceiling; three $30M apps can beat one $90M app
- Mosaic Group: Launch a new app for a new use case instead of piling features into oneGuest says slim apps segment users better and can earn more; no numbers
- RoboKiller: After the base subscription, offer data broker removal as a paid add-onReally successful; a good way to raise revenue per user (no numbers)
- Mosaic Group: Count only 2-4 years of revenue per user and require fast payback early onConservative at worst; high LTV:CAC shows how much more spend cash could unlock
- Mosaic Group: Before rolling out a price test winner, check it still wins if retention drops 10-50%Annual-heavy price calls sometimes proved wrong a year later; this guards against it
- Mosaic Group: Raise prices about 10%; retention usually drops only a couple of percentNet LTV wins most of the time across the portfolio, if conversion holds
- Mosaic Group: Use early subscription cancellations to predict who will churnDidn't work: few cancel early; most cancel near the renewal reminder
- RoboKiller: Track whether each user is getting the core value, and step in when you fail themA wrongly handled call was highly correlated with churn, triggering fixes
- RoboKiller: Tell users how many millions of spam calls you blocked for everyone this month
- RoboKiller: Publish your own spam call data so reporters and regulators cite youRepeated WSJ, FT and cable news mentions; a brand halo that drove many organic installs
- Mosaic Group: Move ad designers to a different app every couple of monthsMore varied creatives, so winning styles became clear; no numbers
- Launch one ad channel at a time and change one bid per monthGuest's post-ATT method; new channels almost always underperform early forecasts
- Apalon: Switch apps from one-time paid downloads to in-app subscriptionsPhenomenal growth after the switch (no numbers)