What is HR analytics, and how do you build it in Metabase?
HR analytics turns recruiting, hiring, workforce, and people-ops data into shared metrics leaders can act on. In Metabase, you build it by syncing ATS and HRIS data into a database, modeling candidates, applications, requisitions, workers, and organizations, then publishing permissioned dashboards.
TL;DR — ATS data powers recruiting funnel, time to hire, time to fill, offer acceptance, and source-quality dashboards. HRIS data powers headcount, org, location, employment-status, and time-off dashboards. Metabase reads the database you sync these tools into; it does not connect directly to HR SaaS APIs.
What does HR analytics measure?
- How fast are we filling roles, and where does the process stall?
- Which candidate sources produce qualified applicants, offers, and accepted hires?
- Are we on track against the hiring plan by department, location, and role family?
- How is headcount changing by organization, manager, status, and location?
- Where do people operations need attention, without turning analytics into surveillance?
Which tools feed HR analytics?
| Tool | Best for | MCP for AI-assisted analysis |
|---|---|---|
| 100Hires | AI-assisted ATS: jobs, candidates, applications, interviews | Registry-listed official remote server |
| Greenhouse | Structured ATS and recruiting operations | Community and gateway Harvest API wrappers |
| Ashby | ATS, CRM, scheduling, and recruiting analytics | Gateway MCP wrapping Ashby ATS API |
| Workday | Enterprise HCM, organizations, positions, requisitions | Community Workday HCM servers |
| BambooHR | SMB HRIS, employee records, time off, hiring workflows | Community and gateway BambooHR API wrappers |
What is the shared HR analytics data model?
Build these as clean models before broad dashboarding:
modeled_jobsormodeled_requisitions— one row per role or opening.modeled_applications— one row per candidate-role relationship.modeled_application_stage_history— one row per stage/status event.modeled_interviews— interview schedule, outcome, feedback status.modeled_offers— offer extended, accepted, rejected, start date.modeled_workers— employee or worker record from HRIS, permissioned tightly.modeled_organizations— department, location, manager, cost center.
Which HR metrics matter most?
- Time to hire — candidate/application start to accepted offer.
- Time to fill — requisition open to accepted offer or start.
- Candidate conversion rate — stage-to-stage funnel movement.
- Offer acceptance rate — accepted offers divided by extended offers.
- Source quality — qualified outcomes by source, not just applicant counts.
- Interview pass-through rate — candidates who advance after an interview stage.
How do you connect HR tools to Metabase?
Metabase reads SQL databases, warehouses, and uploaded tables. Use a managed connector, dlt, API job, approved report export, or Workday RaaS to sync HR data into a database, then connect Metabase to that database. MCP can help with quick AI-assisted lookups, but recurring HR dashboards need history and permissions.
Which dashboards should you build first?
- Recruiting funnel — applications, stages, conversion, aging, and drop-off.
- Hiring plan — open roles, accepted offers, starts, plan vs. actual, roles at risk.
- Source quality — qualified candidates, interviews, offers, accepted hires by source.
- People analytics — headcount, departments, locations, employment status, and time off.
Common mistakes
Averaging time to hire or time to fill.→ Hiring durations are right-skewed. Use median and p90, and segment by role family or seniority before drawing conclusions.
Calculating conversion without stage history.→ Current status is not enough. Use every application stage event to know who reached each step.
Building one dashboard for every audience.→ Recruiting ops may need candidate-level queues. Leadership usually needs aggregate trends. Use different models, collections, and permissions.
Related
Dashboards
Integrations
Metrics
FAQ
What is the difference between HR analytics and recruiting analytics?
Recruiting analytics focuses on candidates, applications, requisitions, interviews, offers, and sources. HR analytics also includes employees, organizations, headcount, time off, and workforce planning.
Can I join ATS data to HRIS data?
Yes, but only after defining the grain. ATS applications are candidate-role records; HRIS workers are employment records. Join through requisitions, starts, or worker IDs after hire, and respect permissions.
Should candidate notes be in Metabase?
Usually not in broad models. Keep notes and sensitive profile fields out of shared dashboards unless a tightly permissioned operations workflow explicitly needs them.