Gaming
Installs are trivial to buy. Day-30 retention decides whether it mattered.
User acquisition for free-to-play titles, measured on retention curves and payback windows.
+48%
Improvement in day-seven cohort return on ad spend
Mobile user acquisition
The problem
Almost everyone who installs your game will never spend anything. Revenue is a power law, the median player is worth zero, and you are effectively buying probability of a high-value player rather than buying transactions. That would be manageable if you could observe the outcome, but on iOS you largely cannot: the signal comes back aggregated, delayed, capped at six bits of conversion value, and nulled entirely when cohorts fall below Apple’s privacy thresholds.
Then the payback window makes it a finance problem. A mid-core title recouping over ninety to a hundred and eighty days is not running a marketing line item, it is running a working-capital cycle. Getting the ROAS curve wrong is not a bad quarter, it is a cash position.
Considerations
What we account for
Treat creative as the media buy
Networks bid on your behalf. The genuine levers are creative, event mapping and target ROAS — and of those, creative moves the most. We run to a production cadence with installs-per-mille tracked by concept and by variant, refresh thresholds set in advance, and a distinction maintained between new concepts and iterations on a winning one. Playables are treated as a build with their own production track, not as a video edit.
Design the conversion value schema on purpose
Six bits of fine conversion value is a design constraint, and most schemas are inherited rather than chosen. We map what you actually need to distinguish — early monetisation signal, retention milestone, or predicted value bucket — against postback windows and crowd anonymity tiers, accepting that you cannot have all three. Getting this wrong is not recoverable by optimisation later.
Model to the payback window the title actually has
Hypercasual, casual and mid-core recoup on completely different curves, and applying one title’s target ROAS to another is a common and expensive error. We set cohort ROAS checkpoints appropriate to genre and monetisation model, then track actual cohorts against the curve rather than against a blended average that hides the shape.
Reconcile the three numbers that will disagree
Network-reported performance, SKAdNetwork postbacks and your MMP will not agree, by design rather than by fault. We run a standing reconciliation and, where the disputed budget justifies it, settle the question with geo holdouts or matched-market tests rather than by picking whichever dashboard is most convenient.
Services
What we run here
PPC Advertising
Paid search and shopping, managed against margin rather than platform-reported ROAS.
Performance Marketing
Cross-channel budget allocation decided by measured contribution, not platform self-reporting.
Customer Acquisition
Acquisition economics modelled to payback period, then media bought to fit.
Campaign Management
ROI-focused campaign planning, pacing and review run as an operating rhythm.
Frequently asked
How do you actually measure iOS now?
Through SKAdNetwork postbacks with a deliberately designed conversion value schema, moving to AdAttributionKit as coverage allows, plus incrementality testing where the aggregate signal is too coarse to answer the question. What we do not do is present modelled data as though it were observed. When a number is an estimate, the report says so.
How much creative do you need to run this properly?
More than most studios expect, and the volume scales with spend. The useful framing is that creative supply has replaced bid management as the main operational constraint. We agree a monthly concept and variant target with you at the start, because a media plan that outruns the creative pipeline stalls within weeks.
Can you work with a soft launch?
Yes, and it is the right time to involve us. Soft launch is where measurement infrastructure and the conversion value schema should be settled, and where kill criteria should be agreed while nobody is emotionally committed to the outcome. Agreeing what a failed retention curve looks like before you see it is considerably easier than afterwards.
Do you handle ASO as well?
We coordinate it and factor store conversion rate into UA planning, since it directly affects effective cost per install. Dedicated ASO work — keyword research, store page testing, localisation at scale — we would either scope separately or recommend a specialist. It is adjacent to what we do rather than the same thing.
Start with the audit.
It has a defined scope and a defined deliverable, and it is deliberately separable from anything that follows. If the audit says your current setup is fine, that is a legitimate outcome and we will say so.