Instagram Shopping Tab — Measuring Off‑App Purchases, Opportunity Sizing, and Launch Readout

Context

Instagram is planning a new Shopping tab. Users often discover products on Instagram but complete purchases on third‑party sites, which Instagram cannot directly observe. Design a rigorous, end‑to‑end plan to (A) verify the hypothesis that Instagram discovery drives off‑app purchases, (B) size the pre‑launch opportunity given assumptions, and (C) execute a launch readout that localizes funnel drop‑offs, quantifies incrementality vs. cannibalization, distinguishes UX vs. supply issues, and rules out data quality problems.

Part A — Verify the off‑app purchase hypothesis

• What measurable in‑app signals would you use as proxies for off‑app purchase intent? Examples include outbound merchant‑link clicks, profile → bio link navigations, saves/shares with commerce hashtags, and search‑after‑view patterns.
• How would you quantify potential bias (e.g., selection, novelty, confounding from seasonality) and construct counterfactuals (e.g., synthetic controls, matched cohorts)?

Part B — Pre‑launch opportunity sizing

• Assumptions: DAU = 50M, 10% show intent weekly (by a defensible in‑app proxy), each intentor averages 0.2 purchases/week off‑platform, AOV = $45, future platform take rate = 5$ 0.002 per impression for Shopping‑tab surfaces with an estimated 6 impressions per intentor per week.
• Estimate weekly gross revenue impact by component: transaction fee vs. ads.
• What sensitivity analyses would you run, and which 3 assumptions most dominate the estimate?

Part C — Launch readout and investigation

You launch and observe high click‑through into merchant flows but purchases far below forecast. Build a metric tree and investigation plan that:

1. Localizes drop‑offs across the funnel (Discovery → Product View → Deep Link → Cart → Purchase).
2. Evaluates cannibalization of existing commerce surfaces (e.g., Marketplace) vs. true incrementality (design a geo or user‑level holdout).
3. Distinguishes product/UX frictions from supply/quality issues (relevance, price, shipping).
4. Rules out instrumentation/data‑quality problems. Specify primary success metrics, guardrails (e.g., session length, feed CTR, ad revenue lift, complaint rate), required segmentation cuts, and what evidence would trigger a rollback.