An iOS game’s mediation strategy should maximise the revenue earned from available ad opportunities while protecting player retention. With less granular targeting data available, studios need to examine which buyers can value their inventory, whether ads arrive in time to be shown, and how monetisation changes affect lifetime value. In-app bidding can improve competition for impressions, but the result depends on demand coverage, implementation quality and the economics of each placement.
“Post-IDFA” describes an environment with restricted access to the identifier, rather than its complete disappearance. Since iOS 14.5, App Tracking Transparency (ATT) has required permission before an app can access the IDFA or track users across other companies’ apps and websites. Without authorisation, the advertising identifier returns zeros. Apple also makes clear that substituting another identifier does not remove the permission requirement. Apple’s tracking requirements establish the boundaries within which the mediation stack must operate.
The commercial implication is that some impressions carry less information for buyers that previously depended on individual targeting and attribution. Those buyers may lower bids or redirect budgets when they have less confidence in predicted returns. Publishers can consequently face pressure on eCPMs and, where eligible demand becomes insufficient, fill rates. The size and direction of that effect depend on geography, format, audience and demand mix. A studio should establish the impact in its own cohorts before attributing every revenue decline to ATT.
SKAdNetwork addresses campaign attribution through a different mechanism. Its privacy protections limit what advertisers can learn from conversion postbacks, and reporting follows conversion windows rather than providing an immediate record of every user’s activity. In SKAdNetwork 4, crowd anonymity influences the detail returned, including whether conversion information is fine, coarse or absent. These constraints affect the feedback available for campaign optimisation; they do not directly set a publisher’s eCPM or prevent an ad from being served. Apple’s SKAdNetwork explanation describes these distinctions.
Studios should also account for AdAttributionKit, which builds on SKAdNetwork and supports privacy-preserving attribution of installations and re-engagement. Apple states that using the framework for attribution does not itself require ATT permission, although any tracking still does. Confirm support with each advertising partner and attribution provider when planning SDK updates. Apple’s AdAttributionKit overview provides the relevant framework requirements.
Within those constraints, bidding gives participating demand sources an opportunity to price the current ad request. A waterfall instead calls sources sequentially according to configured eCPMs, which may be entered manually or informed by historical performance. When those rankings become stale, they can misrepresent a network’s value for the next impression. Bidding can reduce that mismatch by introducing current offers into the allocation decision. This is a way to improve price discovery with the available signals; it cannot recreate the targeting information a player has declined to share. Google’s mediation guide explains the bidding and waterfall models.
A hybrid configuration can still be useful when a waterfall partner contributes demand that bidding sources do not cover. Its auction mechanics deserve scrutiny. In AdMob, for example, the winning bidder is positioned alongside waterfall sources according to eCPM. A waterfall source ranked above that bid can receive the first opportunity to serve. Keep its configured value aligned with observed results so an inflated estimate does not repeatedly take priority over a live bid. Google documents this hybrid sequence.
Choose partners according to their incremental contribution. Test whether a new source increases total revenue within the same segment, fills previously unserved opportunities, or improves ad availability. Examine the SDK’s effect on startup time, stability and maintenance effort alongside its revenue. A network winning impressions that would otherwise have been sold is not sufficient evidence of an overall gain; its presence must improve the outcome for the game.
Use segmentation to make that evaluation credible. Start with major markets and ad formats, then examine placements and ATT authorisation status where reporting supports it. Compare similar traffic and keep enough volume in each group to make decisions meaningful. An eCPM difference between authorised and unauthorised users may also reflect differences in geography or behaviour. Avoid treating that comparison as a clean estimate of the effect of consent.
Trace performance from the player’s ad opportunity through the mediation request, successful load and displayed impression. Record no-fill responses, timeouts and display failures separately. Check the denominator behind each platform’s reported fill rate before comparing dashboards. A strong load rate can coexist with weak revenue if players leave before the ad is displayed, while a high eCPM can conceal a large number of unserved opportunities.
Price floors need the same scrutiny. Raising a floor can remove lower bids from consideration and improve the average price of impressions that remain, yet reduce total revenue if too much demand is excluded. Test floors by relevant segment and evaluate revenue per opportunity alongside eCPM. Confirm which sources each control affects: AdMob’s mediation-group bidding floor applies to bidders and overrides the ad-unit floor for that bidding traffic. Google’s floor documentation also notes that bidders respond differently to floors.
Consider an illustrative comparison with 1,000 equivalent ad opportunities. A configuration producing 800 impressions at a $12 eCPM earns $9.60. Another producing 950 impressions at an $11 eCPM earns $10.45. The second earns more despite its lower eCPM. These are hypothetical figures, but they show why studios should judge yield using both realised price and displayed volume, then account for fees and any change in player behaviour.
Delivery speed belongs in that calculation. Preload rewarded ads ahead of a likely request, within the network’s cache and expiry rules, and offer the placement when an ad is ready. Instrument loading delays and reward delivery failures so the team can distinguish weak demand from implementation problems. For interstitials, choose natural breaks and define what the game should do when an ad is unavailable. Avoid leaving a player waiting through repeated loading attempts.
Before changing commercial settings, verify SDK and adapter compatibility, ad-unit mappings and initialisation. Google’s iOS mediation guidance specifically advises waiting for initialisation to finish before loading ads so networks can participate fully in the first request. Test the integrated stack across relevant permission states and device conditions. Google’s iOS integration documentation explains this dependency.
Permission handling must remain consistent across the stack. Follow each SDK’s instructions for passing applicable privacy signals and requesting ads under the user’s choices. Apple’s data-use rules hold developers responsible for third-party code in their apps and prohibit device fingerprinting. Contextual signals and internal gameplay analysis must therefore be used within the permitted data practices. ATT authorisation is not blanket permission for every subsequent use of player data.
For revenue analysis, capture impression-level callbacks where supported and retain placement, source, currency and precision information. Google’s SDK distinguishes precise values from estimates and publisher-provided values, so an event labelled as revenue is not necessarily a final settlement amount. Reconcile event totals with network reporting and investigate material gaps before using them to tune floors or forecast LTV. Google’s impression-level revenue documentation describes those precision categories.
Keep publisher revenue measurement and acquisition attribution distinct. Revenue from an ad shown inside the game answers a different question from a postback crediting a campaign for bringing a player into it. For acquisition analysis, design conversion values around meaningful indicators of player value, allow reporting windows to mature, and treat missing detail as uncertainty. A missing conversion value should not automatically be interpreted as zero revenue.
Validate changes with concurrent experiments where feasible. Keep assignment stable, change one major variable at a time, and define the observation period and decision criteria before launch. Use ad revenue per daily active user alongside retention, payer conversion and total revenue per user. A placement change that raises today’s ad revenue can still weaken the game’s economics if it reduces future sessions or purchases. Rewarded-ad participation and completion also deserve attention, particularly when changing the value or frequency of rewards.
Studios seeking help with implementation and ongoing optimisation can evaluate OptAd360’s iOS game monetization services. The company describes Bidlogic as an automation layer working across AppLovin MAX and Unity LevelPlay, and offers support for existing monetisation setups. Define a baseline and agree on reporting access, responsibilities and success measures so performance can be assessed against the game’s own results.
A practical next step is to select one meaningful market and placement, establish its current revenue per opportunity and delivery failure rate, then test a specific change to demand coverage, floors or loading behaviour. Expand the configuration when the evidence shows a sustained benefit within the game’s retention and stability limits. That process gives studio leaders a defensible basis for investing in mediation as targeting and attribution capabilities evolve.
