Advertising and credit-risk decisions jointly shape acquisition cost, delinquency, and collections for a lending app. This article adapts Xiao Wei's 2023 discussion into a shared measurement framework. The source reflects a particular business context; test any approach against your own data and applicable local requirements.
Why do acquisition and risk teams disagree?
Media buyers may acquire cheap installs that rarely become funded customers. Risk teams may tighten approvals after delinquency rises, potentially rejecting viable applicants. Both teams need a shared funnel from ad impression to mature repayment rather than judging the business by one department's metric.
Use one funnel and shared definitions
Track comparable cohorts across impression, click, install, application, approval, disbursement, due date, and repayment. Document event definitions, time zones, attribution windows, and data sources.
- Acquisition: CTR, CPI, application CPA, funded-customer CPA, and traffic quality by market and creative.
- Risk: application completion, approval, delinquency, due-date collection, and outcomes by risk tier.
- Business: mature-cohort net income after acquisition cost and credit losses.
Neither cheap installs nor a high approval rate alone justify scaling. Compare the full cost and collections of the same cohorts.
Choose optimization events by signal quality
The original suggested installs during early learning, applications in the middle stage, and disbursements or value events once risk decisions stabilize. That is a case-specific path, not a universal rule. Consider how closely an event tracks value, its daily volume, reporting reliability, delay, and selection bias from approvals.
Applications are frequent but distant from repayment. Disbursement is closer to value but depends heavily on approval policy. Repayment is closest to revenue yet may be sparse and delayed. See our event and user-quality case study and Google App campaign guide.
A practical collaboration process
- Product, risk, and media teams jointly define valid applications, funded loans, and due-date collections, including data-use boundaries.
- Set acceptable risk and acquisition-cost ranges for each market and product version. Version every approval-rule change with its effective date.
- Test one major variable at a time, such as optimization event or approval threshold; avoid simultaneously changing budget, creative, events, and rules.
- Wait for funded cohorts to mature before comparing approvals, delinquency, collections, and net income. Investigate seasonality and traffic-mix changes.
- Hold a regular joint review of traffic mix, risk tiers, and application-flow issues.
Which source claims need caution?
The original cited fixed approval and delinquency thresholds, app-store ratings, and an assumption that older apps naturally have better risk. These are not validated universal benchmarks. Lending ads, data use, credit decisions, and collections must respect applicable local requirements and customer rights.
For web-to-app measurement, check consent, identifiers, event deduplication, and postback scope before following the current AppsFlyer documentation.
FAQ
Should media or risk lead?
Both own the business outcome and should be accountable for the decisions each controls.
Should we always optimize for repayments?
No. Repayments are valuable signals, but sparse or delayed events may not support stable optimization.
How do we assess ads after a risk-policy change?
Tag the policy version and start date, compare market and funded cohorts, and avoid changing multiple media variables at the same time.