Hyper-casual

How a hyper-casual publisher used LTV and ROAS to stop the bleed and scale

Consider a lean hyper-casual publisher running several titles at once, buying installs across a handful of ad networks. In hyper-casual, the math is brutal and the margins are thin: monetization is mostly ad revenue plus the occasional in-app purchase, retention is short by design, and the whole business lives or dies on whether each dollar of user acquisition comes back fast enough.

The challenge

The team was scaling spend on blended numbers. Their network dashboards each reported their own “ROAS,” computed differently, and none of them tied ad revenue back to the actual cohort of users a campaign brought in. So they were flying blind in two directions at once: they couldn’t tell which channels were quietly losing money, and they couldn’t tell when a campaign was supposed to pay back — D3? D7? Never? When a title’s overall ROAS dipped, they had no way to know whether one bad source was dragging down three good ones.

What they did with Keentics

They started with cumulative LTV. Keentics builds the LTV curve from full event and revenue data — ad revenue and IAP combined — so they could see, day by day, how much a fresh cohort was actually worth over its first two weeks rather than guessing from a lifetime average. That curve became the denominator everything else hung off of.

Next they layered in ROAS payback: cumulative LTV against acquisition cost, read as a curve over days-since-install. Now “is this profitable?” had a precise answer with a date attached — the day the LTV line crosses spend. Campaigns that never crossed were no longer ambiguous; they were just losing money on a clock everyone could see.

The decisive move was segmentation. They split LTV and ROAS by channel, by campaign, and by geo, all on the same cohort definition. The blended picture fell apart in the most useful way: one network looked fine on average but was carrying a cluster of geos that never paid back, while a “boring” channel turned out to have the steepest early LTV of the lot. They cross-checked the lifetime value definition so finance and UA were arguing about the same metric, then acted: cut the bleeding geos, hold the marginal ones, and push budget into the source with the fastest payback.

The result

In situations like this, isolating the losing slices and reallocating toward the fastest-recovering channels tends to lift blended ROAS by a meaningful margin and shorten average payback by a few days — directional, illustrative figures, not an audited result, and the real numbers depend on the titles and the auction. The structural win was a single shared view of LTV and ROAS, segmented the same way for everyone, so “scale or kill?” stopped being a debate and became a reading off a chart.

For a publisher whose edge is reacting faster than the competition, the value wasn’t a clever model — it was seeing channel-level payback clearly enough to move budget with confidence, days earlier than before.

This is a representative hyper-casual scenario rather than a named customer story, but the channel-by-channel LTV-and-ROAS workflow is exactly how disciplined UA decisions tend to get made.