How do you analyze product feedback, surveys, and product experience data in Metabase?
Product and customer insight tools hold the record of what customers ask for, how they feel, and what they actually do: feature requests and votes, NPS and CSAT responses, and usage and frustration signals. In Metabase, bring entities and aggregates from those tools into a SQL database and model three adjacent layers — feedback items, survey responses, and usage rollups — joined on account and segment. Those layers power dashboards for the request pipeline, NPS and CSAT programs, feature adoption, and session frustration, next to the CRM and revenue data the rest of the company already reports on.
TL;DR — Never sync raw event or session streams into a BI warehouse. Sync requests, votes, responses, and daily usage rollups. MCP is useful for live exploration; durable product reporting needs database-backed syncs and history.
Which tools does this cover?
- Pendo — feature and page usage rollups, NPS responses, guide engagement, accounts and visitors
- Productboard — feedback notes, features, releases, objectives
- Aha! — ideas, idea votes, features, initiatives
- Canny — posts (feature requests), votes, comments, boards
- Featurebase — posts (feature requests), upvotes, comments, changelog entries
- Typeform — survey responses, forms and fields, answers, completion events
- SurveyMonkey — survey responses, surveys and pages, collectors, questions and answers
- Qualtrics — survey responses, surveys and questions, distributions, contacts and directories
- Tally — form submissions, forms and fields, answers, respondent metadata
- Fullstory — session and event rollups, defined events, frustration signals (rage and dead clicks), segments
- LogRocket — session rollups, frontend errors, rage and dead clicks, conversion events
- Contentsquare — behavioral event rollups, journeys and funnels, frustration scores, heatmap aggregates
For instrumented event analytics on the same questions, cross-link the shared model to PostHog, Amplitude, and Mixpanel — usage rollups from either side land in the same warehouse tables.
What is the shared product insights data model?
| Concept | Common terms | Used for |
|---|---|---|
| Feedback item | Post, idea, request, note | Pipeline, themes, close-the-loop |
| Vote | Vote, upvote, insight link | Vote velocity, demand weighting |
| Status change | Workflow transition, roadmap stage | Aging, flow, time to ship |
| Survey response | Response, submission, interview | Response and completion rates |
| Answer | Answer, field value, score | NPS, CSAT, drop-off by question |
| Usage rollup | Feature usage, defined event, page rollup per day | Adoption, breadth, engagement |
| Frustration signal | Rage click, dead click, error click | Experience quality by flow |
| Account | Account, company, workspace, visitor | The join key for everything above |
How do you connect these tools to Metabase?
- MCP + CLI route — pull a scoped, summarized export through a tool's MCP server, save CSV, and load it with
mb upload csvfor fast exploration. - Pipeline route — sync requests, responses, and rollups into a database or warehouse with a connector, native export (Pendo Data Sync, Fullstory Anywhere, Heap Connect), or API pipeline.
- Cross-source route — join feedback, survey, and usage data to CRM, support, and revenue tables to explain outcomes, not just collect signals.
What can you analyze across these tools?
- Net Promoter Score — percent promoters minus percent detractors, with response volume
- CSAT — satisfied responses over all satisfaction responses, per touchpoint
- Survey response rate — responses over an explicit denominator
- Survey completion rate — completed over started responses
- Feedback vote velocity — new votes on open requests per period
- Time to close the loop — request created to shipped and announced
- Feature adoption rate — adopting accounts over eligible active accounts
- Rage-click rate — frustration signals normalized per session
Which dashboards should you build?
- Voice of customer overview — scores, requests, and response behavior in one view
- Feature request pipeline — statuses, vote velocity, aging, and time to ship
- NPS & CSAT trends — score trends with volume and segment breakdowns
- Survey performance — response and completion rates, drop-off by question
- Feature adoption & engagement — adoption, breadth, activation, and declining accounts
- Session frustration signals — rage clicks and friction by page, flow, and release
Common mistakes
Syncing raw event or session streams into the warehouse.→ Land entities, votes, responses, and daily rollups. Raw streams belong in the source tools; the warehouse is for trends and joins.
Ranking feature requests by lifetime vote totals.→ Old requests always win. Rank by vote velocity and weight by segment or revenue.
Reporting NPS or CSAT without response volume.→ Score swings on thin samples are noise. Chart volume next to every score and enforce a sample floor.
Comparing rates with inconsistent denominators.→ Response rates over sends, opens, and views differ wildly — standardize one denominator per program and label it.
Counting internal users in adoption and frustration numbers.→ Your own team skews both. Exclude internal domains and test accounts from every headline card.
Related
Analytics
Integrations
FAQ
What is product and customer insights analytics?
Product and customer insights analytics is reporting built on what customers say and do: feature requests and votes from feedback boards, NPS and CSAT from surveys, and usage and frustration signals from product experience tools — synced into a SQL database and analyzed in Metabase. It answers the questions the individual tools can't answer alone: does what we ship match what customers ask for, and does the experience improve when we ship it? See the product feedback analytics overview for the full build.
Does Metabase connect natively to Pendo, Typeform, or Canny?
No — Metabase reads SQL databases and warehouses, not SaaS APIs. Sync each tool's data into a database first with a connector, native warehouse export, or API pipeline, then point Metabase at that database. For a quick first pass, pull a summarized export through the tool's MCP server and load it with the Metabase CLI (
mb upload csv). Each tool guide — Pendo, Typeform, Canny — documents both routes.Why combine feedback, surveys, and product experience data?
Because each source alone misleads. Feedback boards over-represent your loudest customers, surveys capture attitudes but not behavior, and usage data shows what people do but not why. Joined on shared account and segment fields, they cross-check each other: a voice of customer overview next to adoption data shows whether shipping the most-requested feature actually moved usage or NPS.
How do these tools differ from PostHog, Amplitude, or Mixpanel?
Product analytics platforms (PostHog, Amplitude, Mixpanel) are event pipelines you instrument yourself. This cluster covers the tools around them: feedback boards where customers vote, survey platforms that measure sentiment, and session tools that auto-capture usage and frustration without instrumentation. Most product teams run one of each — and the warehouse is where their data finally joins.
Which dashboards should you build first from this data?
Start with the feature request pipeline if feedback runs your roadmap, or NPS & CSAT trends if a survey program is the center of gravity. Add the voice of customer overview once two or more sources are synced, then feature adoption and session frustration signals as usage rollups land.
Which tools does this category cover?
Twelve tools across three groups: product feedback and roadmaps (Productboard, Aha!, Canny, Featurebase), surveys and NPS (Typeform, SurveyMonkey, Qualtrics, Tally), and product experience (Pendo, Fullstory, LogRocket, Contentsquare, which also covers Hotjar and Heap). All map onto shared feedback, survey-response, and usage-rollup models.