LTV, CAC and payback period: how to steer ad spend for subscription apps

ROAS on day one says almost nothing about a subscription app. Payback period and the ratio between LTV and CAC do. Here is how to model those numbers conservatively and set spend limits you can defend.

Scaling a subscription app on ROAS means steering on a metric that does not understand the business model. A subscriber pays out over months or years, while ROAS only counts what lands inside the attribution window. The numbers that actually matter are payback period, the time until a subscriber earns back their own acquisition cost, and the ratio between LTV and CAC. This article covers how to model both without fooling yourself, and how to derive spend limits from them that you can explain to yourself and to your investors.

Why does ROAS not work for subscription apps?

ROAS divides revenue by ad spend within a window of days. For e-commerce, where the customer pays in full at checkout, that is a reasonable approximation of reality. For a subscription app it is a caricature. A new subscriber might pay ten euros in the first month, while their real value depends on how many months they stay. On day seven every campaign looks unprofitable, and based on that picture you switch off campaigns that would have been your best cohorts a year later.

The reverse happens too. A campaign pulling in cheap trials can look beautiful in the dashboard while those trials cancel en masse after the first month. In that case ROAS rewards exactly the wrong behavior: lots of volume, little value.

How do you model LTV without fooling yourself?

LTV is not a number you look up, it is a curve you build from your own data. The foundation is cohort analysis: group subscribers by start month and track per cohort how many are still paying after one, two, three and six months. That retention curve, multiplied by your net revenue per subscriber per month, gives you the cumulative value of a cohort over time. Now you have an LTV based on behavior instead of hope.

Three principles keep the model honest. Calculate with net revenue, so after store fees, payment costs and refunds, because that is the money you actually pay ads with. Extrapolate carefully: young cohorts have no proven tail yet, so cap your projection at a horizon your data supports instead of compounding into infinity. And model per channel and per campaign type, because a subscriber from a trial campaign behaves differently than one who paid straight away.

An optimistic LTV is not a forecast, it is a permission slip to overpay for users.

What does the payback period really tell you?

The ratio between LTV and CAC tells you whether a user eventually returns more than they cost, but not when. That is where the payback period comes in: the number of months until a cohort's cumulative net revenue passes its acquisition cost. Two apps with the same LTV to CAC ratio can have completely different paybacks, and that difference dictates how fast you can scale.

Payback is a cash flow metric. Every euro of ad spend is locked up until the cohort has earned it back. The shorter your payback, the more often you can reinvest the same euro per year and the less outside capital your growth requires. A long payback can be perfectly fine, but only if you have the cash to bridge the months in between. What counts as acceptable is therefore not a benchmark but a function of your own runway and funding.

How do you set spend limits you can defend?

With a conservative LTV curve and a measured payback period, budgeting becomes arithmetic instead of gut feeling. The reasoning we use:

• Set your maximum CAC per channel from your cohort LTV and the margin you want to keep at minimum.
• Set from your cash position how much spend can be outstanding at once in cohorts that have not hit payback yet.
• Raise budgets only when new cohorts prove their retention, not because installs happen to be cheap.
• Recalibrate the model every quarter; retention shifts with every product update and every new audience you open up.

The beauty of this approach is that it settles two conversations at once. Internally it gives your team a clear frame: below this CAC you may scale, above it you may not. And externally it gives investors or your bank a defensible story: not we think it will work out, but this cohort behavior supports this level of spend.

What does this mean for your campaigns?

Optimize your campaigns toward the events closest to value that still give Meta enough volume to work with, and judge campaigns on cohort quality instead of cost per install. A campaign with pricier installs but better retention almost always beats a campaign with cheap installs that drain away after a month. You only see that when you connect your Meta data to your own cohort data, and that connection is exactly what is missing in most app accounts we take over.

Conclusion

For subscription apps the question is never what is my ROAS, but how fast does a cohort earn itself back and how much spend can I justify on that basis. Model LTV conservatively, steer on payback, and you can scale aggressively without blowing up your cash flow. Building paid social on that logic, from event selection to campaign structure to cohort level reporting, is exactly what we do for app businesses. Want to know where your payback stands today and what that means for your spend? Book a call and we will gladly take a look with you.