Build something that is not there yet
Measure attrition and tenure by hire cohort
- When leadership asks whether attrition is rising and where, and the answer takes a week to produce.
- When a hire-cohort retention question can't be answered because leaver events aren't classified consistently.
Build annualized attrition by org and tenure band and a hire-cohort retention triangle, classifying leaver events as voluntary, involuntary, or regretted from the headcount snapshot. Covers the attrition and retention measures, not the exit-interview data behind the classification.
Sample This Recipe has not been materialized in the Cookbook repository yet. Its trigger, description, prompt, agent guidance and acceptance conditions, and the explanation below, are prototype drafts. Its name, job, area and readiness come from the reconciled Cookbook seed snapshot. Readiness is a separate question from this one: it says whether the capability exists, not whether the writing has been reviewed.
Use this Recipe
Use the VibeData Recipe `people-attrition-tenure` at https://getvibedata.ai/cookbook/people-attrition-tenure Read the Recipe and execute it in the context of the current Intent.
Recipe id people-attrition-tenure · Not yet materialized in the Cookbook repository, so the pointer addresses this page.
Verified by
What has to be observably true before this Recipe is finished.
- leaver counts reconcile to the headcount roll-forward's leaver movements for the same period
- the retention triangle is monotonic within a cohort (retention never rises as tenure increases)
- aggregate cells below the agreed minimum group size are suppressed
- a voluntary/involuntary/regretted split sums back to total leavers for every period
Recipe promptThe task specification the agent reads. Reference only — it is not what you copy.
Deliver: Measure attrition and tenure by hire cohort. Build annualized attrition by org and tenure band and a hire-cohort retention triangle, classifying leaver events as voluntary, involuntary, or regretted from the headcount snapshot. Covers the attrition and retention measures, not the exit-interview data behind the classification. Execute inside the current Intent. Its Domain, repository, platform, environment and attached sources are the context for this work — read them rather than asking for them. The work is done when: - leaver counts reconcile to the headcount roll-forward's leaver movements for the same period - the retention triangle is monotonic within a cohort (retention never rises as tenure increases) - aggregate cells below the agreed minimum group size are suppressed - a voluntary/involuntary/regretted split sums back to total leavers for every period Report the evidence for each condition above with the result. A condition you cannot meet is something to say, not something to work around.
Agent guidanceHow the agent approaches the work, and what it will not do.
Profile the inputs the grain, joins and measures actually depend on before proposing a model. Put the design up for review — grain first — then build in an isolated copy with tests and documentation landing beside the model rather than after it.
Composes
- dbt model authoring and layering
- dbt in the project
- isolated-copy execution and gate verification
Asks first
Semantic decisions the Intent cannot supply. Never context Studio already holds.
- which Fabric target this work lands on, when the Domain carries both a Lakehouse and a Warehouse
- the grain the requester expects, where the request leaves it open to more than one reading
Guardrails
- Build in an isolated copy. Production is read, never written.
What you need
- Microsoft Fabric Lakehouse, Microsoft Fabric Warehouse, MotherDuck, or DuckDB.
- A dbt project you can build, and read access to the models it starts from.
- dbt in the project, or the intent to add it.
- Needs the probation window and internal-transfer treatment agreed up front.
How it goes
- State the outcome in one sentence, in the language the request arrived in.
- Let it profile the inputs the grain, the joins and the measures actually depend on.
- Review the design. Disagreeing about grain here costs a sentence; after the model exists it costs a rewrite.
- Let it build in an isolated copy, with the tests and the documentation landing beside the model rather than after it.
- Read the acceptance conditions against the run.
What you end up with
The deliverable, in your own repository, as transformation work a reviewer who knows the project reads as native to it. Alongside it, the evidence for every one of the acceptance conditions above — which is the part that is still there in three weeks when somebody asks.