Context
The Instagram sharing team wants to test a new Share to Story button on feed posts and IG Reels. The product goal is to increase downstream engagement in the AARRR funnel by making it easier for viewers to re-share content into Stories, which may drive more creators, viewers, and re-engagement.
Hypothesis Seed
The proposed treatment adds a more prominent Share to Story CTA next to the existing share affordance on eligible Instagram posts and Reels. The team believes this will increase sharing behavior and, through that, improve overall engagement such as session depth and content interactions (including IG Save) without harming core experience quality.
Constraints
- Eligible traffic: 12M Instagram DAU/day globally on iOS and Android, with ~35% seeing at least one eligible feed post or Reel per day
- Maximum experiment window: 14 days; PM needs a ship decision before the next Instagram app release train
- Randomization must happen in Meta’s experimentation platform at user_id level
- False positives are costly because a noisy UI change could degrade feed quality at massive scale; false negatives are acceptable up to one release cycle
- The team expects only a modest lift, so you should use CUPED with a 7-day pre-experiment covariate if appropriate
Deliverables
- Define the null and alternative hypotheses, including whether the primary test should be one-sided or two-sided.
- Choose a primary metric, 2-4 guardrail metrics, and 1-3 secondary metrics. Be explicit about the metric definitions, unit of analysis, and where IG Save fits.
- Compute the required sample size and minimum runtime using a clear MDE, real numbers, and assumptions about baseline rates. State whether CUPED changes the runtime materially.
- Specify the experiment design: unit of randomization, allocation, stratification, and how you will handle peeking, multiple comparisons, and SRM checks.
- State a pre-registered ship / don’t ship / iterate rule that respects guardrails, and discuss pitfalls such as Novelty Effect (primacy effect), network interference from sharing, and possible SUTVA violations if friends see shared Stories.
