What is sales analytics, and how do you build it in Metabase?
Sales analytics turns the activity in your CRM — leads, deals, stages, activities, and owners — into shared metrics about how your pipeline is built, where deals stall, and how reliably you hit the number. In Metabase, you build it by syncing your CRM into a database, modeling a small set of clean tables (especially a deal + stage-history model), and standing up dashboards anyone can read.
What does sales analytics measure?
It measures the health and productivity of your pipeline — not just a single bookings number. The durable, leader-friendly questions are:
- How much pipeline do we have, and does it cover the target? (pipeline coverage)
- How often do deals we work actually close? (win rate, conversion rate)
- How long does it take to close? (sales cycle length)
- What is a typical deal worth? (average deal size / ACV)
- Where do deals stall in the funnel? (stage-to-stage conversion)
Avoid vanity metrics (raw activity counts, total leads ever created). They're easy to grow and rarely change a decision.
Which tools feed sales analytics?
The same pattern applies to every CRM. Per-tool setup lives on each integration page:
| Tool | Best for | MCP for AI-assisted analysis |
|---|---|---|
| HubSpot | All-in-one CRM for SMB to mid-market | Official remote server (OAuth) |
| Salesforce | Configurable enterprise CRM | Official hosted servers (read-only option) |
| Pipedrive | Pipeline-first CRM for SMB sales teams | Official native server (OAuth) + community |
| Attio | Flexible, data-model-driven CRM | Official remote server (OAuth) |
| Close | Built-in calling and outreach for SMB sales | Managed connector (Pipedream) |
| Zoho CRM | Full-suite CRM with multiple pipelines | Official composable servers (OAuth) |
| Dynamics 365 Sales | Enterprise CRM on Dataverse | Managed connector; Dataverse SQL (TDS) for sync |
| Twenty | Open-source CRM on PostgreSQL | Managed connector; direct DB for self-hosters |
What is the shared sales data model?
Almost every CRM maps onto these entities. Model them as clean tables, not raw connector JSON:
| Concept | Common terms | Used for |
|---|---|---|
| Lead / contact | Lead, contact, person | Funnel entry, sourcing |
| Account | Company, organization, account | Segmentation, ABM |
| Deal / opportunity | Deal, opportunity | Pipeline, win rate, ACV |
| Stage | Deal stage, pipeline stage | Conversion, funnel |
| Stage change | Stage history, field history | Sales cycle, velocity |
| Activity | Call, email, meeting, task | Effort, engagement |
| Owner | User, rep, owner | Attainment, workload |
The single most important field is a reliable stage-change history. With it you can compute true sales cycle length, stage-to-stage conversion, and velocity. Without it, those metrics must be caveated — a point-in-time snapshot can't reconstruct when a deal entered or left a stage.
Which sales metrics matter most?
Define each one once and reuse the definition everywhere:
- Win rate — deals won ÷ deals closed (or created) in a cohort. Fix the denominator and the cohort before you compare.
- Sales cycle length — created → closed-won. Report the median, not the average — the distribution is heavily right-skewed.
- Pipeline coverage— open pipeline ÷ quota or target for the period. Needs an accurate open-pipeline snapshot.
- Conversion rate — stage-to-stage and overall funnel conversion; shows where deals advance and where they stall.
- Average deal size— won amount ÷ won deals (ACV). Watch mix shifts and outliers; report the median alongside the mean.
Once you bring in the tools around the CRM — sales engagement and revenue-intelligence platforms — a few more metrics come into play: reply rate, meetings booked, email bounce rate, and forecast accuracy. Join those signals back to the CRM opportunity so they map to real pipeline.
How do you connect a CRM to Metabase?
Two complementary routes, the same for every tool:
- MCP route (AI-assisted) — pair the tool's MCP server with the Metabase MCP server for live, exploratory questions. Treat it as exploratory, not governed reporting, and remember it creates no history — so it can't reconstruct stage-change velocity or historical pipeline on its own.
- Pipeline route (warehouse-backed) — sync the tool into a database with a managed connector (Airbyte, Fivetran) where one exists, or with dlt / the tool's API, then build durable dashboards on a deal + stage-history model.
Which dashboards should you build first?
- Pipeline — open pipeline by stage, owner, and segment, with coverage against the target.
- Forecast — weighted and unweighted pipeline for the period, with commit vs. best-case.
- Win rate — won ÷ closed by cohort, segment, source, and stage entered.
- Sales activity — calls, emails, and meetings by rep, framed as effort and coverage rather than surveillance.
- Lead funnel — stage-to-stage conversion and drop-off from lead to closed-won.