Source Contribution Measurement System
Measure whether a source adds useful people—not whether it can return a lot of profiles.
Last reviewed: 2026-09-06
Ask SourcingOS about this page
Inputs
- Frozen role artifact
- Lane/source-level discoveries
- Identity-resolved candidates
- Evidence-fit retention decisions
- Provider cost/credit data where applicable
1. Preserve raw discoveries by source
Record what each source found before global ranking or result caps hide later-source contribution.
2. Resolve identities across sources
Deduplicate carefully so the same person found through several sources does not inflate total supply or suppress unique contribution through false merges.
3. Apply one evidence-fit threshold
Compare sources against the same role requirements and retention threshold rather than allowing one provider to count weaker profiles as successful discoveries.
4. Calculate retained and unique contribution
Report raw discovered, evidence-fit retained, duplicate retained, and unique retained counts. Add latency and cost/credits where useful.
5. Segment by role family and geography
A source can be excellent for engineering and weak for healthcare, or strong in one geography and poor in another. Avoid universal rankings without representative coverage.
What to measure
- Raw discoveries
- Evidence-fit retained yield
- Unique retained yield
- Duplicate pressure
- Cost per unique retained candidate
- Latency to usable result
Failure modes to avoid
- Comparing headline database size
- Counting duplicate profiles as source value
- Using different fit thresholds by provider
- Ignoring cost and latency
- Declaring one universal provider winner from one role