Why boomerangs moved up the agenda
Every labour market cycle produces a hiring channel that was always available and only becomes a programme when the alternatives get expensive. Boomerang recruiting — rehiring former employees — is that channel for the current cycle. The economics are unusually good: a rehire arrives pre-vetted on culture and standards, ramps faster than an external hire at the same level, carries known performance history, and in most firms carries lower acquisition cost than agency-sourced equivalents. Post-exit experience frequently adds exactly the skills the firm lacked — alumni who left for competitor firms or adjacent industries return with structural knowledge you cannot buy on the open market.
Most firms' rehire activity, however, is emergent rather than managed: a hiring manager reconnects with an ex-colleague, HR processes the return, and the event is recorded — if at all — as a new-hire row with no connection to the prior tenure. The difference between emergent and managed is measurement, and measurement is a data-infrastructure problem before it is a programme. This piece is about the infrastructure.
The measurement problem: what even counts as a rehire
Ask three HR teams for their rehire rate and you will get three numbers built on three definitions. The definitional decisions that move the number:
– Separation window. Does someone who leaves and returns after three weeks count as a rehire, or is that a payroll artefact? Common practice sets a minimum gap (90 days or more); whatever the choice, it must be fixed and documented, or the metric drifts with each reporting analyst.
– Entity scope. Firms with multiple brands, subsidiaries, or acquired entities see cross-entity moves that one system calls a transfer and another calls an exit-and-return. Decide at group level.
– Denominator. Rehires per total hires is a mix metric; rehires per eligible separations is the programme metric. The first can rise while the programme does nothing — it just means external hiring fell. Report the second.
Then there is the censoring trap that every duration metric inherits. Time-to-rehire computed only over people who did return is biased short — you are averaging over survivors. The honest treatment is survival analysis: every exit enters the denominator at exit, returns are events, and the clock is reported as a curve with censoring at the observation boundary ("of the 2021 exit cohort, X% had returned by year three"), not a mean over the lucky. A CFO-facing number built the other way will be quietly wrong in the flattering direction, and quietly wrong in the flattering direction is how programmes get over-funded and then cancelled.
The substrate: what the data layer must actually hold
A measurable rehire programme consumes four data properties, and they are precisely the properties a browsable directory does not provide:
– Employer-verified exit records. Who worked here, in which roles, with effective dates, and how the exit was classified. This is HRIS-derived, not member-declared — it is the difference between "someone claims they worked here" and "the payroll system says they did, until March 2023, as a director." Every downstream metric is anchored on this.
– Current-role freshness. A rehire pipeline is a sourcing system, and sourcing needs to know what alumni do now. Current-role data comes from consented profile updates, enrichment where lawful, and engagement signals — and it decays continuously. A programme that cannot see that its best boomerang candidate was promoted at a competitor last quarter is not running a pipeline; it is running a mailing list.
– Engagement and consent signals. Which alumni want career contact, which want only event invitations, which opted out entirely. Outreach to non-consented alumni is both a privacy problem and a brand problem — the alumni network is read by current employees, and an ill-targeted "come back!" wave becomes the watercooler story about HR's database. Consent state is a first-class field in the pipeline, not a checkbox in the privacy annex.
– Exit classification with nuance. Good-leaver status, exit reason, and eligibility flags — some exits are permanently ineligible (for-cause terminations, non-solicit-restricted moves), and some are the most valuable targets (high performers who left for exactly the experience they now bring back). Without classification, the pipeline sprays; with it, the programme prioritises.
Why directory-only tools cannot carry this
General community platforms and lightweight alumni portals hold member-declared profiles, event RSVPs, and forum posts. They are not connected to the HRIS, so they cannot anchor on verified employment; they have no exit classification because they never saw the exit; they have no rehire event loop because the return happens in a system they do not touch. The employer-verified substrate — import pipelines from HR data, deduplication, effective-dated history — is the unglamorous architecture that makes the alumni base a graph rather than a guestbook, and it is the reason the enterprise category's platform consolidation has happened around record-centric vendors rather than feed-centric ones. If your current tool cannot answer "how many 2022 engineering exits are now at competitor firms," it is a directory, and your boomerang programme is running on top of it by faith.
The metrics that survive audit
– Rehire rate: rehire events per eligible separation, rolling 24-month cohorts, fixed 90-day-plus gap definition, group-level scope.
– Time-to-rehire: survival curves by exit cohort, censoring disclosed, reported as "returned by year N" percentages.
– Quality of rehire: 12- and 24-month retention of rehires versus external hires at matched levels — the number that defends the channel's premium, and the one that must be tracked from the first cohort because it cannot be reconstructed later.
– Source-of-hire discipline: alumni-sourced hires attributed through an actual source field, with the "direct/unknown" bucket shrinking each year. If the unknown bucket is large, the programme's best hires are being credited to nobody.
Operating notes before the first outreach wave
Three governance decisions precede the metrics: honour non-solicit and notice obligations with explicit eligibility flags rather than case-by-case judgment; keep outreach inside consent boundaries even when the CRM makes it frictionless; and remember that the alumni audience overlaps the candidate audience overlaps the current-employee audience — every boomerang email is also a signal to the people who stayed about what leaving and returning looks like at this firm. Run the programme as infrastructure — verified records, honest definitions, audited metrics, consent at the centre — and the returns compound quietly. Run it as a campaign, and the only number it produces is a send count.