Mobile

Hybrid Monetization ROAS: How to Combine Ad Revenue (ILRD) and IAP Signals

Most mobile games earn from ads and in-app purchases. Learn how ILRD and IAP revenue reach your MMP, Google, Meta and SKAN as one honest ROAS signal.

Hardal
7 min read
Ad revenue and in-app purchase revenue joining into one ROAS signal for a mobile game

Most mobile games no longer earn from one source. A player watches rewarded videos for a week, then buys a starter pack. Another never pays but sees hundreds of interstitials. If your campaigns only learn from in-app purchases, they ignore most of your players. If they only learn from ad revenue, they miss your best payers. Hybrid monetization needs a hybrid revenue signal, and that signal has to reach every system that bids for you.

This post explains where each kind of revenue comes from, how it reaches your MMP, Google, Meta and Apple's attribution frameworks, and where server-side data helps.

What ILRD is, and why it matters for ROAS

ILRD means impression-level revenue data. Your mediation platform (AppLovin MAX, Unity LevelPlay, AdMob and others) reports a revenue value for each ad impression, along with the ad format, ad unit and network that filled it. Because the value arrives per impression on a known device, you can tie it to a user and then to the campaign that acquired that user.

Without ILRD, ad revenue only exists as a daily total in the mediation dashboard. You can see that ads earned money yesterday, but not which campaign brought the players who earned it. That is why ad-heavy games often look unprofitable in ROAS reports that only count purchases.

Why IAP and ad revenue tell different stories

The two revenue streams behave differently, and a blended number hides that unless you plan for it.

  • Precision. An in-app purchase has a store price and a store receipt. Many ILRD values are estimates. AdMob, for example, marks each value with a precision type: PRECISE, ESTIMATED, PUBLISHER_PROVIDED or UNKNOWN, and reports values in micros. Keep the precision field, so you know how much to trust each number.
  • Timing. A purchase is one event at one moment. Ad revenue builds up across many small impressions over days and weeks.
  • Reversals. Purchases can be refunded. Apple sends a REFUND notification through App Store Server Notifications V2, and Google Play exposes refunds and chargebacks through the Voided Purchases API. Ad revenue has no refund event, but estimates can differ from the final payout.
  • Fraud and testing. Device-side purchase events can be spoofed unless they are validated against the store. Test impressions can show up with a value of zero.

None of this is a reason to avoid blending. It is a reason to keep both streams labeled, so each destination gets the version it can use.

Which system needs which signal

Each system that bids or reports on your game reads revenue in its own way.

DestinationHow it reads revenueWhat to watch
Your MMPIn-app events plus ad revenue, through the SDK, mediation integrations, and its server-to-server APIEvent naming, currency, and one event ID per purchase
Google Ads App campaignsGoogle Analytics for Firebase events. Ad revenue uses ad_impression with value and currencyGoogle says bidding events for App campaigns should come from the Firebase SDK
MetaApp events with value and currency, from the SDK, the MMP, or the Conversions API for app eventsDeduplicate the SDK and server copies of the same purchase
SKAdNetwork and AdAttributionKitA conversion value schema you define: a fine value from 0 to 63 in the first SKAN 4 postback, and coarse values (low, medium, high) in all threeMap revenue ranges so both IAP and ad revenue can move the value

Two details deserve attention. First, Google offers tROAS for ad revenue, which optimizes toward ad_impression revenue. For hybrid games, Google lets you add ad_impression to a tROAS campaign next to your in-app purchase events. If your mediation platform is not AdMob, you log ad_impression yourself from its ILRD callback. Second, SKAN conversion values are chosen on the device inside short windows, so ad revenue only counts if your schema includes it. Our SKAdNetwork guide covers postback windows and conversion values in more detail.

Where server-side data fits, and where it does not

The device is still the right place to capture an impression or start a purchase. Some revenue truth, however, only exists on a server. Store validation, renewals and refunds come from Apple and Google. Web shop purchases happen outside the app. Final ad payouts come from network reporting.

A first-party data layer can join those sources and send corrected revenue to your MMP through its server-to-server API. In practice that means validated purchases instead of raw device events, refunds sent as negative or reversal events, web shop sales tied to the player, and a shared event ID so each purchase is counted once. Our event deduplication entry explains why that ID matters.

Be honest about the limits. Server-side delivery does not override App Tracking Transparency or a consent choice. It cannot create attribution where the user opted out. For Google App campaigns, the bidding events still come from the Firebase SDK, so the server's role there is validation and reconciliation, not replacement.

A checklist for a hybrid revenue signal

  1. List every revenue source: IAP, subscriptions, ad revenue by mediation platform, and any web shop.
  2. Confirm that ILRD is captured for every ad format, including impressions with a value of zero, and that precision and currency are stored.
  3. Validate purchases against the store and send refunds from the server.
  4. Use one event ID per purchase across the SDK and the server.
  5. Decide which revenue each destination should optimize toward, and write it down.
  6. Check that your SKAN or AdAttributionKit schema lets ad revenue move the conversion value.
  7. Each month, reconcile ILRD totals against network payouts and note the gap.

If you already send revenue signals to your MMP, our earlier post on why your MMP needs revenue signals, not just installs is a good companion. For how MMPs and ad platforms each claim the same install, read Web ad, app install: who gets the credit?.

FAQ

What is ILRD in mobile games?

ILRD stands for impression-level revenue data. A mediation platform reports an estimated revenue value for each ad impression, which lets you attribute ad revenue to a user and to the campaign that acquired them.

Is ILRD the same as the ad revenue I get paid?

No. Many ILRD values are estimates. Final payouts come from the networks' reporting later, so use ILRD for optimization and cohort ROAS, and reconcile it for finance.

Can Google Ads App campaigns optimize toward ad revenue?

Yes. tROAS for ad revenue uses the Firebase ad_impression event with value and currency, and hybrid campaigns can combine it with in-app purchase events.

Can server-side tracking recover revenue from users who declined tracking?

No. It follows the same consent and App Tracking Transparency rules as the SDK.

Do we still need an MMP?

Yes, for most studios. A first-party data layer improves the revenue the MMP receives. It does not replace attribution, SKAN or AdAttributionKit handling.

Start with one comparison

Pick one week and one campaign. Compare the revenue your MMP attributes to it with what your purchase records and mediation reports show for the same players. If ad revenue is missing, or refunds never arrive, that gap is where your bidding is learning the wrong lesson.

To see how Hardal joins IAP, ad revenue and refunds into one signal for your MMP and ad platforms, explore Hardal for gaming.


Hardal is the first-party data and agentic analytics platform. We help enterprises send accurate signals to their ad platforms and grow revenue, using server-side data from their websites, mobile apps and AI agents.

FAQs
5 Questions

Frequently Asked Questions

Direct answers on ILRD, hybrid ROAS, Google tROAS for ad revenue, and where server-side data fits next to an MMP.

What is ILRD in mobile games?
ILRD stands for impression-level revenue data. A mediation platform such as AppLovin MAX, Unity LevelPlay or AdMob reports an estimated revenue value for each ad impression, usually through an SDK callback. That value is what lets you attribute ad revenue to a user, and so to the campaign that acquired them.
Is ILRD the same as the ad revenue I get paid?
No. Many ILRD values are estimates. AdMob, for example, labels each value with a precision type (PRECISE, ESTIMATED, PUBLISHER_PROVIDED or UNKNOWN). Final payouts come from the networks' own reporting later, so ILRD is good for optimization and cohort ROAS, and needs reconciliation for finance.
Can Google Ads App campaigns optimize toward ad revenue?
Yes. Google offers tROAS for ad revenue, which uses the Google Analytics for Firebase ad_impression event with value and currency. For hybrid games, Google lets you add ad_impression to a tROAS campaign alongside in-app purchase events. Google states the bidding events for App campaigns should come from the Firebase SDK.
Can server-side tracking recover revenue from users who declined tracking?
No. Server-side delivery can make revenue events more complete and more consistent, but it must follow the same consent and App Tracking Transparency rules as the SDK. It is not a way around a user's choice.
Do we still need an MMP?
Yes, for most studios. The MMP handles attribution, SKAN and AdAttributionKit postbacks, and its ad network integrations. A first-party data layer next to it improves the revenue it receives, through the MMP's server-to-server API.