D7 ROAS is the first return number most UA teams look at on a new campaign, and it is also the one most often misread. This post explains what D7 ROAS means for mobile games, how to judge it against your payback goal, and what we change when it comes in low.
What is D7 ROAS?
D7 ROAS is the revenue a group of players has earned by day seven after install, divided by what you spent to acquire them. If a day’s installs cost 1,000 and those players have earned 250 by day seven, D7 ROAS is 25 percent.
Three details matter:
- It is cohort based. You follow players who installed on the same day, from the same campaign, rather than everyone playing this week.
- It is cumulative. Day seven includes everything earned from day zero to day seven, not just revenue on day seven.
- It needs complete revenue. For most of our games revenue is about 80 percent ads and 20 percent in-app purchases, so a number built on purchases alone will look far worse than the truth.
A D7 ROAS of 25 percent does not mean the campaign loses money. It means the campaign has earned a quarter of its cost back after a week, and the real question is what happens next. Our post on what a good ROAS looks like covers the difference between cohort and blended numbers in more detail.
How do you read D7 ROAS against payback?
D7 is a forecast, not a result. You use it to predict whether a cohort will reach break-even by the date you care about, often D30, D60 or D90.
The method is simple:
- Pick your payback window. How long can you wait to get spend back?
- Find your curve. Look at older cohorts and note what share of their payback-window revenue had arrived by day seven.
- Set the D7 target. If cohorts usually earn about a third of their D60 revenue by D7, and you need to break even by D60, your D7 target is about a third of the spend. Your own history sets the number, not a benchmark.
- Compare each new cohort to that line. Above it is on track. Below it needs an explanation.
The shape of the curve differs by game. An ad-led game tends to earn early and flatten, so D7 is a strong signal. A game that relies on purchases from a small share of players often earns more slowly, so a modest D7 can still pay back later. That is why the same number can be good for one title and worrying for another. The wider picture is in our guide to LTV for mobile games.
What can mislead you about D7 ROAS?
Before changing a campaign, check these:
- Small cohorts. A handful of installs means one purchase can swing the number. Wait for enough players before you act.
- Late revenue reporting. If ad revenue arrives from the network after a delay, the cohort looks weaker than it is for the first few days.
- Day-of-week effects. Cohorts that install at weekends can behave differently from weekday ones.
- Cutting too fast. A campaign that looks weak at D7 can still recover. Turning off every campaign below target on day seven can remove ones that would have paid back.
We pair D7 with a later checkpoint and with retention before making a big move. Our guide to CPI, IPM, LTV and payback shows how these numbers fit together.
How do you improve D7 ROAS?
ROAS has two sides: what you pay and what players earn. Every lever below moves one or both.
Creative
Creative decides who installs. An ad that shows the real game attracts players who like it, and those players tend to stay and earn. An ad that over-promises can bring cheap installs that leave in the first session and drag D7 down.
We test hooks first, because the opening seconds decide most of the result, and we judge a creative on the revenue of the players it brings in, not only on cost per install. When a creative tires, performance drops gradually, so a steady supply of new variants protects ROAS. Our creative testing framework sets out the weekly cadence we use.
Bids and targets
Networks buy the players your goal tells them to find. If the goal is installs, you get install volume. If the goal is a ROAS target, the network looks for players likely to earn.
Moving from an install goal to a return goal usually lowers volume and raises quality. The sensible way to do it is in steps, with a target close to what the campaign already delivers, then tightening as data builds. Setting an aggressive target too early starves the campaign. We cover how this works at scale in our post on scaling Applovin campaigns.
Events and optimisation signals
Networks can only optimise for what they can see. If purchase and ad revenue events are missing, delayed or wrong, the network is optimising blind.
Check that the events you value most are being sent back to each network, that revenue values are correct, and that the optimisation event is one that fires often enough to give the network something to learn from. For a hybrid game, that usually means passing ad revenue as well as purchases.
Geos
The same game earns different amounts in different countries, and costs differ too. A country with cheap installs can still lose money if its players earn little, while a pricier country can pay back faster.
We compare countries on D7 ROAS and on payback, not on install cost alone. Split campaigns by country so each one can have its own target and budget. Then move budget towards the countries whose cohorts track above your line.
The game itself
UA cannot fix everything. If players leave in the first session or never reach the point where ads or offers appear, no bid change will lift D7. Poor retention is a product signal. Check the first session before you blame the campaign.
A simple order of work when D7 is low
When a campaign comes in below target, we go through the same sequence:
- Confirm the cohort is large enough and revenue is fully reported.
- Check retention. If it is weak across every campaign, look at the game.
- Compare creatives inside the campaign and pause the clear laggards.
- Compare countries and move budget.
- Review the optimisation goal and target, and adjust in small steps.
- Recheck at a later checkpoint before cutting the campaign.
Doing it in this order stops us from changing five things at once and not knowing which one worked.
What should you take from this?
D7 ROAS is an early warning light, not a verdict. Set a target from your own curve, read it with cohort size and retention in mind, and change one lever at a time. Judge the whole system on blended ROAS over a longer window. Across 40 months of aggregated results, our blended ROAS has stayed above break-even, and the habit behind that is to react to trends, not to a single week of data.
Book a free 30-minute call. We'll cover:
- A read on your current CPI, retention and ROAS
- Where we'd look first, and what we'd test
- An honest answer on whether we can help