Most features are noise. Astra tells you which ones print money.
You'd think this requires a product analyst and a 6-week study — Astra cross-joins Stripe and PostHog in 30 seconds and gives you the receipts.
Astra answers "which feature makes the most money" by joining Stripe customer revenue with PostHog feature events at the user level. She maps every paying customer to the features they actually used in their first 30 days, then ranks features by attached revenue (sum of MRR for users who touched the feature) and by lift (revenue ratio of users who used it vs didn't). Output is a ranked table: Feature → Users → Attached MRR → Lift Ratio → Verdict (Killer / Important / Nice-to-have / Dead Weight). She flags the 2 features that justify their existence, the 3 you could cut without losing a dollar, and the 1 that everyone touches but doesn't predict retention. You stop building based on what users *say* they want and start building on what actually drives money.
Stripe API: every paying customer, their MRR, signup date, and tenure. Filter to active subscriptions only.
PostHog: every feature event in the first 30 days per user. Group by feature flag or event name. Need ≥5 distinct events to count as 'used'.
For each feature: sum MRR of users who used it. Compute lift = avg MRR of users-who-used / avg MRR of users-who-didn't. Lift > 1.5 = real signal.
Killer = high attached MRR + high lift + high retention. Dead Weight = low usage + low lift. Vanity = high usage + zero lift (everyone touches it but it doesn't predict revenue).
Lark message with the ranked table, the 2 features to double down on, the 3 to consider sunsetting, and the 1 'vanity feature' that's misleading you.
A ranked table of every feature with attached MRR, lift ratio, and verdict — plus a clear list of features to invest in, hold, or sunset.
Monthly, plus on-demand before every roadmap planning session.
We'll spin up your workspace, hand the prompt to Astra, and you see the answer in 60 seconds. Free.
Try this with AstraClear answers about wallet credit, usage, subscriptions, and how Tycoon charges for work.
Astra flags features with <30 days of usage as 'too early to call' instead of including them in the kill list. She tracks them in a watchlist and re-runs the analysis once they hit the data threshold (typically 30-60 days after launch).
She'll do a coarser analysis using whatever event data exists (Mixpanel, GA4, your own events table). Without per-feature events you'll get plan-tier-level insights (which plan makes the most money) instead of feature-level. Still useful, just less surgical.
Yes. Tag any feature as 'strategic — exclude from sunset analysis' and Astra still reports its numbers but won't recommend killing it. Useful for compliance features (SAML, audit logs) that don't drive revenue but unlock enterprise deals.
First time: ~2 minutes (Astra pulls 6 months of Stripe + PostHog and builds the join). Repeats: ~20 seconds (incremental). The Lark report fits in one screen so you can read it in 60 seconds before standup.