Talent SummonerCandidate Pipeline Conversion Calculator
Calculate recruiting pipeline conversion with explicit counts, denominators, missing-data rules, synthetic scenarios, sensitivity checks and stop conditions.
A candidate pipeline conversion calculator shows how unique candidates move between defined recruiting stages during a stated period. It does not predict replies, interviews, offers, hires, quality, speed or return on investment. Use one role or segment, one process version and one cutoff. Synthetic data are not benchmarks, targets or Talent Summoner results.
Define the event
Write the question first: “What share of unique candidates who reached each stage progressed after that stage was closed?” Record role, cohort dates, cutoff, stage definitions and owner. The public SHRM recruitment conversion article, published 20 September 2017 and checked 4 September 2026, describes funnel measurement from attraction/application through screening, offer and acceptance. It is a methodology reference, not a benchmark.
Use events that exist in the team's records. A possible map is discovery record (S0) -> evidence reviewed (S1) -> outreach approved (S2) -> response recorded (S3) -> review-ready hand-off (S4). Define each entry and exit event, whether it is candidate- or activity-level, who records it, and what closes it. A ranked result, drafted message or inferred interest is not a stage event. Do not label a stage “hire” unless that historical event is separately defined.
Choose the denominator
Conversion is a ratio between a shared cohort and stage definition, not everyone ever seen in an ATS or sourcing tool.
| Measure | Denominator | Numerator | Use |
|---|---|---|---|
Closed conversion r_s | Final dispositions at stage s | Progressed to s+1 | Closed-subset process measure |
Drop-off d_s | Final dispositions at s | Closed without progression | Diagnostic count/rate, not cause |
Closure coverage k_s | Included candidates entering s | Closed at s | Resolution by cutoff |
Progressed-to-date p_s | Included candidates entering s | Progressed by cutoff | Snapshot while records remain open |
Cumulative R_{0:k} | Included candidates entering S0 | Reached S_k | Direct observed path |
Let n_s be included unique candidates entering stage s after exclusions; g_s progressed, x_s closed without progression, o_s open, and e_s raw entries excluded as duplicates, withdrawals or out-of-scope records. Then n_s = g_s + x_s + o_s, and the raw audit is n_raw_s = n_s + e_s. An excluded ID must not also appear in x_s.
c_s = g_s + x_s; r_s = g_s / c_s; d_s = x_s / c_s; k_s = c_s / n_s; and p_s = g_s / n_s. Report r_s as closed-subset conversion when o_s > 0; it is not a final cohort rate. If n_s or c_s is zero, leave the affected rate blank. Missing event history, denominator or closure status is unknown, not zero.
For a consistent closed cohort, R_{0:k} = reached S_k / entered S0. With closed sequential stages and no re-entry, check it by multiplying r_0 x r_1 x ... x r_{k-1}. Never multiply rates from unlike periods, roles, sources or process versions. Count an approved unique-candidate re-entry once.
Worksheet and formulas
Record cohort dates/cutoff, role, process version, IDs, n_s, g_s, x_s, o_s, e_s, target q_k, export/query, owner and verification date. Keep an ID-level audit trail. Calculate at full precision and round displayed percentages. For a fully closed cohort, Required S0 records = CEILING(q_k / R_{0:k}); this is a scenario, not a supply forecast. If R_{0:k} is zero or unknown, show unresolved data and no requirement. Pool counts, not percentages; unlike definitions stay separate.
Synthetic scenarios
For 240 synthetic discovery records, transitions of 96, 36, 15 and 6 give 40.0%, 37.5%, 41.7% and 40.0%. Direct cumulative conversion is 6 / 240 = 2.5%; the product check is approximately 2.5%. A planning target q_4 = 12 yields CEILING(12 / 0.025) = 480 upstream records.
If 100 entered a stage, 60 progressed, 15 closed without progression and 25 remained open, c_s = 75, closed-subset conversion is 80.0%, closure coverage 75.0%, and progressed-to-date 60.0%. Do not publish 60.0% as completed. If dispositions or IDs cannot be verified, the rate is unknown and calculation pauses.
Sensitivity and stop controls
Change one documented input at a time and keep the base case visible. With target 12, changing the 37.5% stage to 50.0% gives 3.3% cumulative and 360 upstream; changing it to 25.0% gives 1.7% and 720. Changing the 40.0% final stage to 50.0% gives 3.1% and 384; changing it to 25.0% gives 1.6% and 768. These are synthetic sensitivities, not recommended targets. Use dated comparable cohorts, show counts, and do not invent confidence intervals or industry comparisons from small samples.
Stop publication or planning use when events are vague or changed; a denominator is zero, missing or target-derived; open records are treated as final; IDs or withdrawals cannot be reconciled; populations differ; counts fail n_s = g_s + x_s + o_s or rates leave 0–100%; provenance is missing; or anyone asks for a reply, hire, quality, speed or ROI claim. Assign an owner and restart date. Do not rank people or groups from one small rate.
For one role, freeze stages, export IDs, reconcile counts, run sensitivities and document the next action and refresh date. Keep offers separate.
Where Talent Summoner fits
Talent Summoner supports candidate sourcing and ranking. Candidate sourcing can provide discovery inputs and candidate ranking can organise supplied material, but the team defines stages, verifies evidence, deduplicates, records outreach and replies, closes dispositions and decides. A ranked result is not a conversion, reply, interview, offer or hire. Use the assumptions tool -> current pricing only for a separately recorded cost assumption; do not turn price into a rate, time-saving or ROI claim.
What is a candidate pipeline conversion rate?
The share of candidates with a final disposition at one defined stage who progressed to the next. Keep numerator, denominator and cutoff visible. It is a process measure, not a hiring prediction.
What denominator should I use?
Use unique candidates with a final disposition: progressed plus closed-without-progression. If records are open, show closure coverage and label the result closed-subset conversion.
What if a stage has no data?
Show the entering count and write unknown when the event history, denominator or closure rule is missing. Use zero only for a verified measured zero; never divide by zero.
Should I compare a benchmark?
No universal benchmark is used. Definitions and timing vary. Prefer dated comparable cohorts and show counts beside percentages.
Does Talent Summoner calculate our conversion?
No. It can support sourcing and ranking; your team owns stage definitions, deduplication, outreach, replies, dispositions, records and hiring decisions. Document any product-cost assumption separately through current pricing.
Copy the worksheet, resolve material unknowns with an owner and source date, and refresh when rules or cutoff change. Keep conversion separate from outcome claims.
