How reporting proves creative wins

The metrics ladder#

Read creative in order: impressions show delivery, taps show angle strength, page views confirm routing, and conversion per page proves persuasion. Skipping rungs misleads: strong taps with weak conversions signal curiosity without proof, while weak taps with strong conversions signal great pages starved of traffic. Cost per tap and cost per install sit at the top as efficiency checks, never as creative verdicts alone. Apple own ad variation docs describe the reporting cuts available per variation and page.

Start from the Apple Ads hub for the family view, with search results reporting showing where taps begin.

Angles versus pages#

Every creative result blends two effects: the angle that earned the tap and the page that earned the install. Compare variations landing on the same page to isolate angle strength, then compare one angle across pages to isolate page strength. Changing both at once produces wins nobody can repeat, so hold one steady while testing the other. Log which element changed per test cycle, because memory blurs within a quarter and libraries lie without notes.

The ad variations guide structures angle tests, while CPP setup structures the page side.

Segment by source and market#

Split reads by traffic source and market before judging: search results visitors arrive with intent while browse visitors need persuasion, and one market's winner often flops in another. Product pages serve 90 markets, unavailable mainland China, so segment investment where reachable revenue lives. Custom page traffic should be read per page against the default, never blended into account averages that hide page level truth. Build the split into dashboards from day one rather than reconstructing it after arguments start.

Source thinking follows the discovery and brand split, with market proof in the 91 markets guide.

Directional cases, honest labels#

Vendor write ups share directional examples such as Hallow style seasonal creative pushes where tailored pages and angles moved results with the calendar. Treat every such figure as directional vendor data, never as a promise for your account, and copy the method rather than the numbers: matched pages, distinct angles, dated seasonal variants. Facetune style page iterations and HelloFresh style seasonal mappings belong in the same honest category: useful patterns, not guarantees. Label vendor numbers as estimated or directional in every deck that cites them.

Confirm benchmark discipline with estimated vendor data habits from the benchmarks guide, and Apple own custom product page docs for what pages can carry.

Cadence that compounds#

Run creative reporting on a fixed rhythm: weekly checks for breakage and spend anomalies, monthly reads for angle and page verdicts, quarterly reviews for library prunes. Weekly verdicts on thin traffic produce churn, while annual reviews let losers spend for seasons; the monthly middle holds. Each cycle should retire at least one loser, promote at least one winner and open at least one fresh test, keeping the library in motion. Tie reporting dates to billing cycles so finance sees creative as managed spend.

Feed verdicts into keyword choices via the CPP keyword guide so terms follow proven proof.

From reads to decisions#

Turn every read into a named decision with an owner and a date: promote, retire, localise, retest, or hold for season. Promote winners into broader traffic and localised siblings, retire losers promptly to free variation slots, and retest near ties with materially different challengers. Hold seasonal winners as dated templates for their return berth rather than deleting them. Decisions without owners decay into debates that reconvene monthly without progress.

When reads disagree with instincts, ask our team for an independent creative audit that follows the numbers.

Frequently asked questions#