This simulated case examines a mobile-games portfolio that keeps tuning advertising while player lifetime value remains flat. Intake can reveal sharp early churn, inconsistent experiment results and several AI ideas competing for attention. Those signals do not prove that retention is the binding constraint. Stage 2 would test the hypothesis using level funnels, churn patterns, consent-aware measurement, stability data and controlled simulations. The directional decision is FIX measurement first, SCALE the proven retention playbook, and conditionally START AI-assisted sequencing on one title. Generative content remains at proof of concept until format and quality controls are demonstrated.
What does this case show? The highest-value AI opportunity may sit in early retention and content sequencing rather than in another round of ad tuning.
A mid-sized free-to-play mobile game developer operating a portfolio of casual puzzle titles. Revenue is mainly advertising-based, with an underdeveloped in-app purchase component. Organic growth has slowed, making lifetime value the constraint on profitable acquisition.
How should the studio increase lifetime value across the portfolio, and which AI initiative should receive the next investment?
The studio sits between content creation upstream and ad demand plus platform distribution downstream. The levers most under its control are content sequencing, first-party measurement and portfolio-level player flow.
The following items may be reported during the initial intake or interviews. They are not treated as measured performance values.
Initial intake, interviews and limited document signals may make the following hypothesis a priority. It is not presented as a confirmed root cause until it is tested against operational evidence.
The working hypothesis is that early retention and fragmented measurement may be the binding constraints on lifetime value, while monetisation tuning is the visible symptom. Difficulty walls may push players out before they build attachment, and weak event and consent data may prevent the studio from verifying any intervention. The priority may therefore be to make measurement trustworthy, industrialise level rebalancing and only then test AI-assisted sequencing.
The following are value hypotheses, not achieved or validated outcomes.
Customer value: Players receive a smoother early experience with fewer artificial frustration spikes. Ad load and content difficulty can be managed around long-term experience rather than short-term extraction.
Internal value: Every point of retained audience compounds into more content exposure, more ad opportunities, more purchase eligibility and a larger profitable acquisition envelope.
Treat ad-frequency optimisation, AI level sequencing and generative level creation as one broad innovation programme.
The AI role, human decision boundary, data readiness and measurement logic are designed together. No scale decision is made without pilot evidence.
This table is not an investment decision or a definitive ranking. Priority and evidence readiness are reassessed during Stage 2.
| Opportunity hypothesis | AI role | Value connection | Evidence readiness | Directional priority |
|---|---|---|---|---|
| AI-assisted difficulty sequencing | Predict + Recommend + Simulate | Improve early retention and lifetime value | Level funnel, cohort and experiment data required | High; conditional pilot |
| Early churn-risk detection | Predict + Explain | Target onboarding and difficulty interventions | Reliable player-event data required | High; analytical foundation |
| Portfolio cross-promotion optimisation | Recommend | Improve player flow across titles | Identity and consent constraints must be addressed | Medium-high; after measurement fix |
| Generative level creation | Generate | Increase content supply | Format and quality control unproven | Hold at proof of concept |
| AI ad-frequency caps | Predict + Recommend | Potential ad-yield improvement | Requires clean controlled experiment | Re-test separately |
The design below is not a solution commitment. It is a pilot hypothesis to be validated.
Retention-First AI Portfolio
The following are readiness hypotheses that require document and process review.
This plan is adapted to data access and client scope. Stage 1 alone does not include implementation or outcome validation.
| Period | Purpose | Main actions |
|---|---|---|
| Days 1–30 | Measurement and stability foundation | Assign the experiment owner, audit consent and event coverage, define the test standard and stabilise titles near platform thresholds. |
| Days 31–60 | First validated retention intervention | Run a pre-registered early-level resequencing test on one flagship title and re-test the ad initiative under the same standard. |
| Days 61–90 | Playbook and AI pilot | Codify the rebalancing playbook, apply it to a second title and launch a bounded AI-assisted sequencing pilot with explicit kill criteria. |
The bounded sequencing test improves retention and lifetime-value indicators while stability and experience guardrails hold.
The model finds the right problem areas, but the recommended sequence or experiment design needs refinement.
Measurement remains unreliable, stability risk is unresolved or the pilot does not beat the manual rebalancing baseline.
Transferable lesson: The best AI investment is not necessarily the most visible feature. It is the intervention tied most directly to the constraint that governs lifetime value.
Assess the available evidence, business value and implementation conditions together.