Research runs
The run model, client research versioning, and why CRs are versioned.
A research run is a batch job that researches one or more clients against a single offering. It is the primary unit of work in Client Path.
Run lifecycle
A run moves through these statuses:
| Status | Meaning |
|---|---|
| Queued | Run has been created and is waiting for workers to pick it up |
| In progress | Workers are actively researching clients |
| Completed | All client research is done |
| Failed | One or more clients encountered an unrecoverable error |
What a run contains
Each run records:
- The offering version and methodology version active at the time it started.
- The company profile version active at the time it started.
- A list of client research records, one per client included in the run.
- A rollup report once all clients are complete: a cross-client summary with cohort analysis.
Client research (CR)
A client research (CR) is the per-client unit within a run. Each CR contains:
- Findings: discrete pieces of intelligence extracted from public sources.
- A report: a prose narrative synthesising the findings.
- A propensity score: 0-100, derived by evaluating findings against the methodology rubric.
- An outreach draft: an AI-written message anchored to the top findings (if enabled).
Versioning and immutability
Client research records are versioned. Once a CR completes, its findings, report, and score are fixed. If you re-research a client (start a new run including that client), a new CR record is created and the old one is preserved.
This versioning is intentional:
- It lets you compare how a client's profile has changed across runs.
- It means findings and scores in reports are always traceable to the exact run inputs (offering version, methodology version, company profile version).
Concurrent runs
You can have multiple active runs at once. For example, you might run one batch against your "Enterprise" offering while another runs against "SMB". Each run is independent.
Run scope
A single run can include as many clients as needed. Very large batches (hundreds of clients) are queued and worked through progressively.
See also: findings, propensity scoring, rollup, starting a run.