On this page
- How D7 ROAS is calculated
- Which revenue counts: IAP, IAA, and subscriptions
- Why CTV's D7 ROAS reads lower than Meta's
- Is CTV failing if its D7 ROAS trails Meta?
- D7 ROAS vs D30 ROAS, ROAS, iROAS, and payback period
- How to set a D7 ROAS target from payback
- When a D7 cohort is ready to read
- Which event to optimize CTV toward
- Subscription apps: trial-to-paid and first-renewal ROAS
- D7 ROAS with Starti AI DASH
- Frequently asked questions
D7 ROAS is the revenue a group of users brings in during its first seven days after installing an app, divided by the ad spend that acquired them. App teams use it as the first read on whether a campaign's users will pay back, weeks before day-30 or day-90 data exists.
The calculation for one install cohort:
D7 ROAS = revenue from an install cohort through day 7 ÷ spend on that cohort × 100
On connected TV, the number needs more care than on Meta or a mobile ad network. A CTV install usually follows an ad seen on the TV, so it is credited through view-through attribution in a short window unless the viewer scans a QR code or follows a link, and part of the revenue lands after day 7. Read alone, CTV's D7 ROAS leaves out installs that arrived outside the window or went to a click on another channel.
How D7 ROAS is calculated
An install cohort is every user who installed on the same day. Its D7 ROAS adds the cohort's revenue through day 7 and divides it by the spend that bought those installs. A cohort of 2,000 installs bought for $10,000 that earns $1,500 by day 7 has a D7 ROAS of 15%.
Measurement partners count day 7 in different ways:
- AppsFlyer: its cohort report states that "Cohort day 0 is the install day," and cohort days follow the calendar rather than the install timestamp.
- Adjust: its cohorts run in 24-hour blocks from the install, with "D0 = 24 hours after install or reattribution."
- Singular: its cohorts use calendar days by default, with rolling 24-hour cohorts available through support.
- Unity: its ROAS guide defines D7 ROAS as "The ROAS generated by a cohort of users within seven days of install."
AppsFlyer notes that its calendar-day cohorts can differ from networks that count from the install timestamp, so an ad network and the MMP can report different D7 ROAS for the same users. Name the tool and its day rule in every report, and compare channels inside one tool. Time zones matter too: AppsFlyer switches a cohort report to UTC when it mixes apps set to different time zones.
Which revenue counts: IAP, IAA, and subscriptions
Revenue in D7 ROAS comes from up to three streams:
- Purchases (IAP): AppsFlyer displays "the exact revenue as sent from the SDK," without app store commissions unless the developer deducted them, and reports net revenue in a separate field. Adjust reports gross IAP revenue and offers an IAP revenue mode that applies a margin to every revenue metric.
- Ad revenue (IAA): in-app ad revenue comes from the mediation platform or ad network and, in AppsFlyer, "is attributed to the media source that originally brought the user," according to its ad revenue guide.
- Subscriptions: trial starts, conversions, and renewals reach the MMP through App Store server notifications and Google Play real-time developer notifications, as Adjust's subscription setup and subscription measurement pages describe.
Each stream matures at its own speed. In AppsFlyer's State of App Monetization 2026, built on revenue from January 2025 through March 2026, in-app advertising had earned 89% of its day-60 revenue by day 7, in-app purchases 60%, and subscriptions 52%. For an ad-monetized game, D7 ROAS is close to the final answer, while for a subscription app it holds about half of day-60 revenue.
Settle gross or net before setting a target. In the US, Apple's standard commission is 30%, or 15% for developers in its Small Business Program, and its subscription terms raise the developer's share from 70% to 85% of the price after a subscriber's first year of paid service. On Google Play in the US, UK, and EEA, the service fees in force since June 30, 2026 are 10% on auto-renewing subscriptions and, for other in-app purchases, 20% on new installs and 25% on existing installs, or 10% on a developer's first $1 million of annual earnings. Each comes with a 5% billing fee when the user pays through Google Play's billing system. CTV drives installs on both iOS and Android, so the share a cohort keeps is each store's share weighted by its iOS and Android mix.
Why CTV's D7 ROAS reads lower than Meta's
CTV reaches the MMP on different terms from a click-based channel:
- Views: a CTV install is credited to an ad seen on the TV unless the viewer scans a QR code or follows a link, and AppsFlyer's cross-platform links credit views for 24 hours by default.
- Priority: Adjust's CTV measurement places CTV impressions last by default, giving them "the lowest priority in Adjust's attribution waterfall," so an eligible click on another channel takes the install. AppsFlyer offers equal priority for CTV impressions and clicks as an optional setting, and the defaults across MMPs are compared in The Evolving Future of Performance TV.
- Delay: a viewer who installs after the window closes is counted as organic or under another channel.
- Household match: the TV and the phone are linked through the household's shared IP address, a probabilistic match covered in CTV ad measurement.
Is CTV failing if its D7 ROAS trails Meta?
Not on that evidence alone. Under Adjust's default, a click on Meta or a mobile ad network takes priority over a CTV impression, so the two D7 numbers are measured on different terms. Compare each channel with its own target, then test CTV against a holdout.
Incrementality testing with a household holdout answers what D7 ROAS cannot: how much revenue the ads added. Google Ads defines incremental ROAS as "Incremental conversion value / Total ad spend," and the reading against attributed ROAS is set out in CTV performance metrics. Read the two side by side. A CTV campaign at 9% attributed D7 ROAS whose holdout shows 12% iROAS over the same weeks caused more revenue than the MMP credited to it; a campaign at 15% attributed and 6% incremental did the opposite. Base budget allocation between CTV and other channels on iROAS, and moves inside the campaign on attributed D7 ROAS.
D7 ROAS vs D30 ROAS, ROAS, iROAS, and payback period
Each metric answers a different question:
- D7 ROAS: an install cohort's first-week revenue ÷ its spend. Shows whether the campaign is on track, read once the cohort's day 7 has matured.
- D30 ROAS: the same cohort through day 30. Shows whether slower revenue, such as renewals and later purchases, is arriving, read after day 30 matures.
- ROAS: attributed revenue ÷ ad spend over any period, which AppsFlyer's cohort report computes as "Revenue / cost * 100." Fits weekly and monthly reporting.
- ROI: return on investment subtracts the cost first, "(Revenue - cost) / cost * 100" in the same report, so break-even sits at 0% ROI and 100% ROAS.
- iROAS: the incremental lift in revenue that a holdout or lift test credits to the ads ÷ ad spend. Shows whether the ads caused the revenue, read when the test ends.
- Payback period: in AppsFlyer's words, "the time it takes to recoup investment in marketing & advertising." Marks the day cumulative net revenue matches spend, which shows how long the cohort takes to pay back.
How to set a D7 ROAS target from payback
Work backward from the day the business wants its spend returned:
- Payback day: the day by which a cohort should return its spend in net revenue, for example day 180.
- Revenue curve: from past cohorts, the share of day-180 revenue that arrives by day 7, for example 25%.
- Net target: 100% × 25% gives a 25% net D7 ROAS.
- Gross target: divide by the share of revenue kept after store fees, weighted by the cohort's iOS and Android mix, the same logic as the "1 ÷ margin" break-even ROAS in Unity's guide. For in-app purchase revenue split 60% iOS at 70% kept and 40% Android new installs at 75% kept, the blended share is 72%, and a 25% net target becomes about 35% gross.
- CTV curve: build the curve from CTV's own cohorts, since its revenue mix and attribution differ, and recheck the target against iROAS once the holdout reads.
When a D7 cohort is ready to read
Day 7 data keeps arriving after day 7:
- Ad network: Unity's IAP ROAS documentation gives January 10 as the earliest maturity date for the D7 cohort of a January 1 install date.
- iOS: for installs measured through SKAdNetwork or AdAttributionKit, Apple's frameworks send conversion data in three windows that close on days 2, 7, and 35 after first launch, with a random delay of "24-144 hours for the second and third postbacks," so day-7 data from iOS can arrive as late as day 13.
- Refunds: AppsFlyer's store revenue documentation says refunds "Can be reported until 180 days after the original purchase," which lowers a cohort's revenue after the fact.
Read D7 ROAS on a week of daily cohorts at a time, once the last cohort in that week has matured and the week holds enough purchasers to be stable, and move bids or targets on that read. For scale, Unity requires 75 unique purchasers with matured day-7 data before a D7 IAP ROAS campaign exits learning. Learning thresholds for Unity and TikTok value campaigns, and Google's rules for changing targets, are covered in the mobile game marketing strategy guide.
Which event to optimize CTV toward
Optimization events, from the most frequent to the closest to revenue:
- Installs: the most volume and the fastest feedback, with no signal on value.
- Registrations or first sessions: fewer events, and a filter for installs that never become users.
- First purchase: closer to revenue, with fewer conversions per day.
- Purchase value: bids toward predicted revenue, once enough purchasers have matured for the model to learn.
Value bidding needs purchase volume first, the way Unity's 75-purchaser threshold does, so move down the list once the next event arrives steadily, and keep D7 ROAS as the read at every step. How to run a CTV campaign covers the first month of optimization in order.
Subscription apps: trial-to-paid and first-renewal ROAS
Revenue from subscriptions arrives in steps. With a 7-day free trial, the first charge falls at the end of the trial, right at the day-7 cut, inside or just outside D7 depending on the tool's day rule and the hour of install. A monthly plan's first renewal comes about a month after that. Three reads fit the steps:
- Trial-to-paid rate: paid conversions ÷ trial starts, read once every trial in the cohort has ended.
- D14 ROAS: first payments from converted trials ÷ spend, read in gross revenue or in net. The developer keeps 70% of the price on the App Store in a subscriber's first year outside the Small Business Program, and 85% on Google Play in the US, UK, and EEA.
- First-renewal ROAS: revenue through the first renewal date ÷ spend, which shows how many subscribers stayed past the first billing cycle.
Example: First-renewal ROAS for a meditation app on CTV
- Cohort: one week of installs, with $20,000 of spend, 2,000 attributed installs, and 600 trial starts.
- Trial-to-paid: 30%, or 180 paid subscribers on a $12.99 monthly plan, read on day 14 after the week's last install.
- D14 ROAS: $2,338 in first payments, 11.7% gross.
- First renewal: 135 of the 180 renew, adding $1,754, for $4,092 and 20.5% gross ROAS, read on day 45 after the week's last install.
- Net view: with installs and revenue split 60% iOS and 40% Android, and first-year subscription proceeds of 70% on the App Store and 85% on Google Play, the cohort keeps 76% of gross revenue, about $3,110, or 15.5% of spend.
- Decision: the team's payback curve for monthly plans expects 18% gross by the first renewal, so the cohort is ahead, and budget grows once the holdout confirms the lift.
D7 ROAS with Starti AI DASH
AI DASH connects CTV exposure to supported visits, installs, registrations, purchases, and revenue through supported web pixels and mobile measurement partners. The Adaptive Performance Engine continuously evaluates campaign performance and automatically adjusts bids and budgets across creatives and media.
Each campaign is measured against its own goal, installs, registrations, or revenue, and the results shape the next round of creative and CTV placement.
Book a demo to plan a CTV test against a D7 ROAS target.
Frequently asked questions
What is D7 ROAS?
D7 ROAS is the revenue an install cohort generates in its first seven days, divided by the ad spend that acquired it, times 100.
How do you calculate D7 ROAS?
Add the revenue from every user who installed on the same day, through day 7, and divide it by the spend on those installs. Check whether the tool counts day 7 by calendar day, as AppsFlyer does, or in 24-hour blocks from install, as Adjust does.
What is a good D7 ROAS?
A good D7 ROAS is the one that puts a cohort on course for its payback day: the share of payback-day revenue that past cohorts earned by day 7, divided by the share of revenue kept after store fees when the MMP reports gross revenue.
ROAS vs iROAS: how does iROAS change the read on D7 ROAS?
D7 ROAS counts all revenue attributed to a cohort in its first week. iROAS, read from a holdout, shows how much revenue the ads caused. On CTV, use attributed D7 ROAS for weekly moves and iROAS to decide whether the budget stays.
Why is D7 ROAS lower on CTV than on Meta?
AppsFlyer's cross-platform links default CTV view-through attribution to 24 hours, and Adjust gives CTV impressions the lowest attribution priority by default, so installs that CTV drove can fall outside the window or go to another channel.
Should in-app purchase and ad revenue both count in ROAS?
For hybrid apps, count both and report them separately. In AppsFlyer's 2026 monetization data, ad revenue reached 89% of its day-60 total by day 7, against 60% for in-app purchases.






