Decision: Tier Comparison Lookback Window
Context
The Guild Hall page shows members “What your tier gets you” and “With [other tier] you would have…” to help them decide whether to upgrade or downgrade. The question: what time window of spending data should power this comparison?
Decision
Use a rolling 30-day lookback window.
The copy reads: “Based on your last 30 days” — transparent, no ambiguity.
Options Considered
Calendar month (rejected)
Useless on April 1st. Members who subscribed mid-month see a partial picture. The boundary is arbitrary relative to their billing cycle.
Billing cycle dates from Stripe (rejected)
Accurate but adds complexity: either an API call at render time (latency, fails locally) or storing billing period dates on the member record (new fields, webhook changes, migration). Over-engineered for the value it provides.
Lifetime average normalized to monthly (rejected)
Fair over time but misleading in month 2 if month 1 was atypical (holiday shopping spike, etc.). Also doesn’t answer the immediate question: “If I upgrade today, what happens next?”
Rolling 30 days (chosen)
Best predictor of the next 30 days is the last 30 days. Answers the member’s real question: “If I switch right now, what would the next billing cycle look like?” Works from day 1 (uses whatever data exists). No external API calls. Simple to explain.
Why This Matters
This feature directly affects churn. If the comparison misleads a member into upgrading when they shouldn’t, or hides value they’re getting, trust breaks. The “no dark patterns” principle means:
- Show honest numbers even if they say “don’t upgrade”
- Use a window that represents real behavior, not a cherry-picked good/bad period
- Label the data source clearly so members understand what they’re looking at
Edge Cases
- New member with < 30 days: Use whatever data exists. “Based on your first 12 days” is still useful.
- Member with no purchases: Don’t show the comparison section at all.
- Backfill month: If someone subscribes and gets backfilled purchases in the same month, the 30-day window includes those at 1x rate. This slightly understates the value of higher tiers but is honest — those purchases did happen at 1x.