Question box
Ask in plain language; Vela maps it to events, cohorts and windows and shows its working. Ambiguous? It asks one clarifying question instead of guessing.
Start with the question. Vela does the rest.
A walk through one real investigation, from the question a PM asked in standup to the answer they shared an hour later, without opening a single dashboard first.
One question, answered
vela · workspace / activation answered 2s ago
ask Why did activation drop last week?
The answer
Mobile sign-ups stalled at the workspace step.
−8.2% activation, week on week
Activation fell 8.2% because 1,940 mobile sign-ups hit the “create a workspace” step and didn’t finish. Web was flat. It started Tuesday, when the step shipped a new required field.
Daily activation
mobile · web
required-field added · v2.41 mobile · Tue 10 Jun
cohort: mobile sign-ups n = 23,418 confidence: 96% view query →
- 01 The read
A verdict, in a sentence
Vela leads with what happened, in language you’d use to explain it to your founder: activation fell because mobile sign-ups stalled at the workspace step. No reading required.
- 02 The proof
The one chart that matters
Not nine charts. The single series that carries the answer, with the change marked and dated, and the rest a click away if you want them.
- 03 The why
The cause, not just the dip
Vela traces the drop to the Tuesday release that added a required field, and shows the step-completion fall that explains it. Cause and effect on the same screen.
- 04 The next move
Pin it, share it, watch it
Set the question to re-run every morning and ping you if it changes. The answer becomes a standing check, not a one-off dig.
Everything Vela does, it does to answer a question.
Cause analysis
When a metric moves, Vela ranks the likely causes by how much each explains, controlling for the obvious confounders, so you debug the number, not your gut.
Release impact
Tie any change to a release and measure the same users before and after, with a confidence range. Know whether the thing you shipped actually did anything.
Cohort retention
Every cohort scored on how it’s likely to retain, with the two or three behaviours that separate the ones that stay from the ones that don’t.
Pre-churn signals
Vela learns each account’s normal usage shape and flags the ones drifting toward churn while you can still reach them, with the behaviour that triggered it.
Drop to raw
Every answer links to its exact query, cohort and event stream. Open the SQL, edit the cohort, export the funnel. The depth is always one click down.
Built for teams that will ask hard questions
Vela runs on your event data, not a sample of it, and is built to be inspected. The answer layer is the convenience; the warehouse-grade engine underneath is the product.
- Ingest
- Stream events via SDK, warehouse sync or the HTTP API. Backfill from your existing pipeline.
- Model
- Bring your own identity resolution and cohort definitions, or use Vela’s. Versioned, so a metric means one thing.
- Query
- The question box, a visual builder, and full SQL over the same engine. Nothing is locked behind the summary.
- Trust
- Every answer carries its query, cohort, sample size and confidence. Reproducible, auditable, shareable.
See Vela answer a question from your data.
Connect a source in a few minutes and ask the first thing on your mind.