Bring back the users you already paid for.
Segments built from the events users actually completed, not from a lookalike model guessing on your behalf.
Segments
Four audiences worth paying to reach twice.
Installed, never activated
Users who opened the app once and never reached the first meaningful action. The cheapest group to recover, and the one most campaigns ignore.
Dropped mid-funnel
Registered but never deposited, browsed but never checked out, started onboarding but never finished it. Messaged against the exact step they abandoned.
Lapsed payers
Users with a completed purchase who have gone quiet past their normal repeat window — usually the highest return per impression on the whole plan.
High-value lookalikes
Built from your actual payers rather than your installers, so the model optimises toward revenue instead of volume.
How it runs
The rules we hold ourselves to.
Retargeting is where performance marketing is easiest to fake. These four constraints are how we keep it honest.
- Segments are built from post-install events your attribution partner already records — no new SDK, no separate tracking layer to maintain.
- Frequency is capped per user per week, because the fastest way to lose a lapsed payer permanently is to follow them around.
- Creative is matched to the abandoned step, so a user who dropped at deposit does not get an install ad.
- Incrementality is checked with a holdout group before we recommend scaling — retargeting is the easiest place in performance marketing to take credit for organic returns.
On incrementality
A returning user would sometimes have come back on their own. We hold out a share of every retargeting audience and report the difference, so the number you see is lift we caused — not activity we happened to sit next to.
Start on a test budget. Scale on the numbers.
Send us your target cost per event and the actions that matter. We’ll run a pilot against your attribution partner and scale only what clears.