Balancing D7 and D28 IAP ROAS: A Game Developer's Guide to Unity Vector Optimization

In the hyper-competitive mobile game development space, User Acquisition (UA) is no longer a simple game of finding users who install and make immediate in-app purchases (IAP). The traditional focus on short-term conversion windows, specifically the Day 7 (D7) cohort, has long been the standard. However, optimizing solely for D7 creates a blind spot: it ignores the long-tail monetization dynamics that define modern free-to-play (F2P) games.

A significant portion of a game's revenue is often generated by high-value players, commonly referred to as "whales" or long-term converters, whose purchasing decisions develop weeks or even months after installation. By ignoring the activity that occurs beyond the first week, studios run the risk of under-valuing high-retention cohorts and over-valuing users who spend quickly but churn immediately. To build sustainable scale, mobile game developers and UA managers must evolve their bidding algorithms.

To address this challenge, Unity Ads, powered by its advanced machine learning framework Unity Vector, has introduced Day 28 (D28) IAP ROAS optimization. This extended window enables algorithms to model player behaviors over a four-week period, allowing campaigns to locate and bid for players who display slower, more deliberate, but ultimately more valuable spending patterns.

Decoupling D7 vs. D28 Optimization Windows

To understand how to best deploy these tools, it is crucial to analyze the mathematical and behavioral differences between the two optimization windows.

The D7 IAP ROAS campaign optimizer relies on predicting a user's cumulative spend within the first seven days post-install. Because the prediction window is short, the feedback loop is fast. The neural network models quickly receive feedback about user conversion rates and ticket sizes, allowing UA managers to adjust bids, budgets, and creatives with high agility. This makes D7 campaigns excellent for rapid testing and for games with aggressive early-game monetization curves.

The D28 IAP ROAS campaign optimizer, on the other hand, extends the predictive horizon to 28 days. The algorithm is trained to recognize subtle, early indicators of long-term engagement—such as session frequency, level completion rates, in-game achievements, and progression speed—and correlate them with a higher probability of late-stage high-value spending. The key advantage is the capability to optimize directly for players with a longer payback period.

Here is a structured comparison of the two optimization windows:

How Unity Vector Predicts Long-Term LTV

At the heart of this optimization is Unity Vector, Unity's proprietary machine learning engine. Rather than relying on simple heuristic bidding (e.g., multiplier-based bidding), Unity Vector utilizes deep neural networks to dynamically predict the expected value of each individual ad impression.

During the install phase, the engine analyzes context signals (geography, device specs, time of day) along with post-install event telemetry. As the user begins interacting with the game, event signals are streamed back to the Unity Ads backend via a Mobile Measurement Partner (MMP) or Server-to-Server (S2S) API. The model processes this sequence of events, using temporal sequence modeling to estimate the player's 28-day cumulative spending curve.

The algorithm then uses this prediction to automatically adjust the real-time bid price. If a user exhibits patterns that strongly correlate with high D28 spending, the engine increases the bid price to secure the impression. Conversely, if a user behaves like an early impulse spender who will quickly churn, the D7 model will value them highly, while the D28 model will de-escalate the bid, protecting the budget from low-LTV spenders.

The Myth of Cannibalization: Why Running Both is Critical

A common concern among UA managers is that running D7 and D28 campaigns simultaneously will lead to audience overlap and self-cannibalization. The assumption is that because both campaigns target in-app spenders within the same game, they will bid against each other in the ad auction, driving up the Cost Per Mille (CPM) and eroding overall ROAS.

However, empirical data and architectural design show the opposite: running both campaigns concurrently creates a complementary effect. This is because they target fundamentally different player profiles.

• The Early Spender: A user who installs a game and immediately purchases a starter pack. This user is highly visible to the D7 campaign model.
• The Late-Game Converter: A user who plays for two weeks for free, deeply engaging with the social and strategic layers of the game, and only makes a purchase during a high-level raid or specialized weekend event. This user is invisible to a D7 optimizer (since their first purchase occurs on Day 14), but is the primary target for a D28 optimizer.

By running both campaigns, you broaden your targeting net. The D7 campaign secures the fast-payback, impulse spenders, providing steady cash flow and quick optimization signals. The D28 campaign acts as a deep-sea net, identifying and acquiring the high-LTV users who form the backbone of the game's long-term economy.

Empirical Performance Benchmarks: Proof from the Field

The performance benefits of utilizing these extended optimization windows are backed by rigorous beta testing and partner case studies. Unity's telemetry across various monetization formats (Ad Revenue, Hybrid, and IAP) indicates substantial uplifts when moving from D7 to D28 models.

In the Ad Revenue ROAS category, beta results showed that 90% of tested games achieved higher D28 retention, with a median retention uplift of +62% compared to D7-only campaigns. Furthermore, 67% of games saw a higher D28 total ARPU (Average Revenue Per User), yielding a median ARPU uplift of +68%.

For Hybrid-monetized titles, which combine in-app ads and IAPs, the gains were similarly impressive. 88% of games recorded higher D28 retention (with a median uplift of +76%), and 59% achieved a higher D28 total ARPU (Average Revenue Per User), a median uplift of +41%.

For pure IAP-driven games, the results are exemplified by mobile publishing giant Homa. In their optimization tests, switching to D28 ROAS optimization resulted in a 14% increase in D28 ARPU alongside a massive 63% increase in player retention when compared to their previous D7 campaigns. These metrics demonstrate that the extended learning window directly translates to high-quality user cohorts who remain in the game longer and spend more overall.

Technical Integration and Event Mapping

To successfully run a D28 IAP ROAS campaign, your game's telemetry pipeline must be configured to supply high-fidelity post-install events to Unity. The precision of the machine learning model is directly proportional to the quality of the data it receives.

UA teams must configure their Mobile Measurement Partner (MMP) or Server-to-Server (S2S) integration to map in-app purchase events with exact transaction values and currencies.

In addition to standard transaction values, passing custom engagement markers (such as level achievements, tutorial completion, guild joins, and daily check-ins) helps the Unity Vector model refine its early predictions before a user makes their first purchase. The model maps these early engagement events to predict the probability of a future transaction, allowing it to bid aggressively on users who have not yet spent money but show high-propensity behavior.

Bidding Strategies and Campaign Setup Workflow

To implement a balanced D7/D28 UA framework, developers and marketing teams should follow a structured deployment workflow:

1. Define the Baseline: Analyze your historical LTV curve. If more than 40% of your total Day 30 revenue is generated after Day 7, your game is a prime candidate for a D28 campaign.

2. Ensure Event Volume: Verify that your MMP is sending all conversion events to Unity. The prediction engine requires a minimum volume of unique purchase conversions per week to complete its learning phase (typically 10-20 unique spenders per geo/campaign group).

3. Setup Parallel Campaigns: Create two distinct campaigns in the Unity Ads console:

• Campaign A: Target: D7 IAP ROAS. Set a realistic target based on historical Day 7 performance.
• Campaign B: Target: D28 IAP ROAS. Set the target based on your projected Day 28 return requirements.

4. Establish Budget Allocation: Allocate 60% of your budget to the D7 campaign initially, as it stabilizes quickly and provides immediate feedback. Allocate 40% to the D28 campaign.

5. Evaluate on Cohort Slopes: Do not compare the two campaigns on a D7 basis. The D28 campaign will look under-optimized on Day 7. Instead, compare their ROAS and ARPU on Day 28. Over time, adjust budget allocations toward the campaign that yields the higher net profit margin at scale.

By decoupling the optimization windows and treating them as distinct, complementary targeting systems, game developers can optimize both immediate cash payback and long-term user retention. Unity Vector’s deep learning capabilities provide the computational backbone, but a structured, data-driven approach to campaign architecture is what unlocks true, incremental scale.