What is product feedback analytics, and how do you build it in Metabase?
Product feedback analytics turns feature requests, votes, survey responses, and product usage signals into shared metrics about what customers want — and whether shipping it moves anything. In Metabase, build it by syncing entities and aggregates from your feedback, survey, and product experience tools into a SQL database, modeling clean feedback, survey-response, and usage-rollup layers joined on account, and shipping dashboards that put the roadmap conversation next to the business data everyone already trusts.
TL;DR — Use three adjacent models: feedback items with votes and status history; survey responses with answers; and daily usage rollups with frustration signals. Keep raw event and session streams in the source tools — the warehouse gets aggregates, entities, and history.
What does product feedback analytics measure?
- What are customers asking for right now — and who is asking?
- Does the roadmap track demand, or drift from it?
- How long do requests wait, and do requesters hear back?
- Are NPS and CSAT moving — and in which segments?
- Do surveys reach people, and where do questionnaires leak?
- Do shipped features actually get adopted?
- Where does the product frustrate users, and did a release cause it?
Which tools feed product feedback analytics?
| Tool | Best for | Getting data into Metabase |
|---|---|---|
| Pendo | Feature adoption and engagement; NPS trend by account segment | Pendo REST API, pipeline, or Pendo MCP server for exploration |
| Productboard | Feature request pipeline by status; Feedback volume by source and segment | Productboard Public API, pipeline, or Productboard MCP server for exploration |
| Aha! | Idea pipeline by workflow status; Vote velocity on open ideas | Aha! REST API, pipeline, or Aha! MCP server for exploration |
| Canny | Feature request pipeline by status; Vote velocity on open posts | Canny API, pipeline, or Canny MCP server for exploration |
| Featurebase | Feature request pipeline by status; Upvote velocity on open posts | Featurebase API, pipeline, or Featurebase MCP server for exploration |
| Typeform | Survey performance and completion; Response volume by form and channel | Typeform Responses API, pipeline, or Typeform MCP server for exploration |
| SurveyMonkey | Survey performance and completion; Response rate by collector and channel | SurveyMonkey API v3, pipeline, or SurveyMonkey MCP server for exploration |
| Qualtrics | NPS and CSAT program trends; Survey performance and completion | Qualtrics Survey Response Export API, pipeline, or Qualtrics MCP servers for exploration |
| Tally | Survey performance and completion; Submission volume by form | Tally API, pipeline, or Tally MCP server for exploration |
| Fullstory | Session frustration signals by flow; Rage-click rate by page and release | Fullstory APIs, pipeline, or Fullstory MCP server for exploration |
| LogRocket | Session frustration signals by flow; Frontend errors by release | LogRocket Streaming Data Export, pipeline, or LogRocket MCP server for exploration |
| Contentsquare | Session frustration signals by journey; Funnel conversion and drop-off | Contentsquare APIs (including Heap Connect), pipeline, or Contentsquare MCP server for exploration |
| PostHog | Instrumented product events, funnels | API, connector, or PostHog MCP |
| Amplitude | Cohorts, retention, adoption joins | API, connector, or export |
What shared data models should you build?
Build clean models on top of raw source tables so each dashboard uses the same definitions.
feedback_items— one row per request or idea, with status, product area, and created/status-changed/shipped timestampsvotes— one row per vote with voter account and timestamp, so velocity and weighting are computablestatus_changes— one row per workflow transition, powering aging and flow analysissurvey_responses— one row per response with started/submitted timestamps and completion flagsurvey_answers— one row per answer with question position, type, and value (including 0-10 NPS scores)product_usage_rollups— daily events and users per feature per accountsession_rollups— daily sessions and frustration signals per page or flowaccounts— the join table: stable IDs, segment, plan, and revenue
Which product feedback metrics matter most?
- Feedback vote velocity — new votes on open requests per period, not lifetime totals.
- Time to close the loop — median time from request to shipped and announced.
- Net Promoter Score — percent promoters minus percent detractors, always with volume.
- CSAT — satisfied responses over all satisfaction responses, per touchpoint.
- Survey response rate — responses over an explicit, labeled denominator.
- Survey completion rate — completed over started, with drop-off by question.
- Feature adoption rate — adopting accounts over eligible active accounts.
- Rage-click rate — frustration signals normalized per session, trended per page.
- Activation rate — signups reaching the value moment, the upstream cousin of adoption.
How do you connect tools to Metabase?
- MCP + CLI — use a tool's MCP server for a scoped, summarized live export, then
mb upload csvfor quick analysis. - Warehouse-backed pipeline — sync requests, votes, responses, and rollups with connectors, native exports (Pendo Data Sync, Fullstory Anywhere, Heap Connect), or APIs for durable dashboards.
- Modeled layer — map each source into the shared feedback, survey, and usage models while preserving tool-specific extension tables.
Which dashboards should you build first?
- Feature request pipeline — statuses, vote velocity, aging, and time to ship.
- NPS & CSAT trends — score trends with volume and segments.
- Voice of customer overview — scores, requests, and close-the-loop rate in one view.
- Survey performance — response and completion rates, drop-off by question.
- Feature adoption & engagement — adoption, breadth, and declining accounts.
- Session frustration signals — rage clicks and friction by page and release.
Common mistakes
Letting lifetime vote totals run the roadmap.→ Totals only grow, so the oldest requests always win. Rank by vote velocity and weight by segment or revenue.
Reporting scores without volume or sample floors.→ NPS on 12 responses is an anecdote. Chart volume next to every score and hold judgment below a minimum sample.
Trying to warehouse raw session or event streams.→ Storage costs explode and queries crawl. Sync daily rollups and entities; keep raw signals in the source tools.
Measuring collection without measuring the loop.→ Requests collected is a vanity number. Track time to close the loop and the share of shipped requests whose requesters were told.
Skipping the account join.→ Without shared account keys, feedback stays anecdotal. Join requests, responses, and usage to CRM accounts so demand can be weighted and outcomes tracked.
Related
Integrations
Terms
FAQ
What is product feedback analytics?
Product feedback analytics is the practice of turning what customers say and do — feature requests, votes, survey responses, and product usage — into governed, shareable metrics. Where feedback tools answer "what are customers asking for?", feedback analytics answers "does our roadmap match demand, do we close the loop, and does shipping move satisfaction and adoption?" In Metabase, you build it by syncing feedback boards like Canny, survey tools like Typeform, and experience platforms like Pendo into a SQL warehouse and modeling shared metrics on top.
What is the difference between feedback analytics and product analytics?
Product analytics (PostHog, Amplitude, Mixpanel) measures behavior from instrumented events. Feedback analytics adds the stated-preference side — requests, votes, and survey scores — and the auto-captured experience side (session frustration signals). The joined view is the point: behavior tells you what users do, feedback tells you why, and the warehouse is where the two finally meet on shared account keys.
Which metrics should a feedback program track?
Start with the loop: feedback vote velocity for what customers want now, time to close the loop for whether anything ships, and feature adoption rate for whether shipped requests get used. Around it, run the sentiment program — NPS and CSAT with response and completion rates — and watch rage-click rate for the experience floor.
Which tools feed product feedback analytics in Metabase?
Any tool whose data lands in a SQL database. This cluster covers 12 guides across three groups: feedback boards (Productboard, Aha!, Canny, Featurebase), survey platforms (Typeform, SurveyMonkey, Qualtrics, Tally), and product experience tools (Pendo, Fullstory, LogRocket, Contentsquare). See the category overview for connection routes.
Is Metabase a replacement for Productboard or Qualtrics?
No — it complements them. Feedback and survey tools are built for collecting, triaging, and acting on individual signals, and they should keep that job. Metabase adds what they don't do well: cross-tool joins, revenue weighting from your CRM, governed definitions, and org-wide sharing — the roadmap conversation next to the numbers the rest of the company runs on.
What data do I need to get started?
Three tables cover the first dashboard:
feedback_items (id, title, status, product area, created and status-changed timestamps), votes (item, voter account, timestamp), and an accounts dimension with segment. That powers the full feature request pipeline. Add survey responses for the sentiment side and daily usage rollups for adoption as the questions arrive.Can I build feedback dashboards without a data warehouse?
For a first pass, yes: pull a summarized export through a tool's MCP server and load it with the Metabase CLI (
mb upload csv) — each upload becomes a queryable table and model. That works for spot-checks. Move to a database-backed sync once dashboards need scheduled refreshes and history — vote velocity and pipeline aging are trend metrics, and a one-off CSV can't trend; every integration guide documents both routes.