Real demand signal from real customer voices. Scored by revenue impact.
You'd think this needs a product manager and a survey — Astra mines every customer touchpoint and ranks requests by revenue impact, not vote count.
Astra mines feature requests from 5 sources weekly and ranks by revenue impact, not just request count. Sources: (1) Intercom conversations tagged or mentioning feature requests, (2) Linear tickets in your 'feature requests' or 'icebox' projects, (3) Granola sales call transcripts where prospects asked for missing features, (4) Twitter mentions of your brand with 'wish you had', 'need', 'add', (5) Cancellation surveys mentioning specific gaps. She clusters into themes via semantic similarity (so 'Notion sync' and 'integrate with Notion' merge), then scores each by: total mentions × average MRR of requesters × % of mentioners who explicitly tied it to renewal/expansion decision. The Lark report ranks top 10 requests with the score, the customer list, sample quotes, and a build-effort estimate. Most teams find their #1 request by vote count is rank 4-5 by revenue impact — and the actual #1 is something a few high-MRR customers have been asking quietly for months.
Intercom conversations, Linear feature-request projects, Granola sales call transcripts, Twitter mentions matching keywords, Stripe cancellation reason surveys.
'Notion sync' + 'integrate with Notion' + 'export to Notion' all merge to one theme. Avoids vote-splitting that hides true demand.
Per theme: total mentions × avg MRR of requesters × renewal-tied % (mentioned in cancellation/renewal context). The score, not raw vote count, is the signal.
For each top-ranked request she pulls existing Linear tickets/RFCs and estimates eng weeks (T-shirt size: S=1wk, M=2-4wk, L=6-12wk, XL=quarter+).
Monday Lark: top 10 requests by revenue impact, customer list per request (so sales/CS can follow up), sample quotes, build-effort tag.
Weekly Monday Lark report: top 10 feature requests by revenue impact, customer list per request, sample quotes, build-effort estimates.
Weekly Monday 9am. Quarterly deep-dive correlating built features with retention/expansion lift.
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.
She'll create a Linear project for you on day 1 and start tagging requests as they come in. For backfill, she scans last 6 months of Intercom conversations and Granola transcripts to seed the database. After 30 days you'll have the full picture even if you've never tracked formally before.
No — she shows both. Vote count tells you what people will tweet about. Revenue impact tells you what they'll pay for. Often they're the same; when they differ, revenue impact wins for build prioritization. The vocal free-tier features can ship later as part of broader releases.
After 6 months of data she can — she correlates 'features shipped' with 'retention/expansion of customers who requested them.' Early on the prediction is rough; by month 6+ she'll tell you 'features in cluster X have an 80% retention lift, features in cluster Y have 0% lift, prioritize X.'
She informs, doesn't decide. The report is data; you weigh it against strategy, technical debt, and team capacity. She'll flag if a roadmap item has zero customer demand signal (a 'we think this is cool' build) — useful for healthy debate, not a veto.