Your Source-of-Hire Report Is Lying by Omission

One person, three touchpoints, three different claims

A candidate first learns about a company from a colleague. Two months later they click an employer-brand post. The final application comes from a tracked job-board link. A report can truthfully label the earliest observed contact “referral,” the latest click “job board” and the self-reported discovery source “colleague.” None is the complete causal story.

Recruiting teams often confuse observed touchpoints with what caused the hire. Tracking links can attribute applications by a stored code. Ashby’s published documentation explains that links may preserve a source in the browser even when the candidate later applies through a job board; the mechanism affects what “source” means in the report. Ashby tracking documentation.

Choose the unit before choosing the winner

One individual can be considered for multiple roles, referred for one and inbound for another. Count applications/job considerations when asking how a specific job’s hiring funnel performed. Count unique people when asking how many distinct individuals you contacted. Count unique starts when asking where accepted hires entered the organization, with a consistent definition for internal transfers.

Vendor documentation explicitly cautions that candidate-level and application-level reports answer different questions because one person can have multiple applications. Avoid adding the same hire twice simply because two applications reached the interview stage. Ashby candidate/application reporting.

Always name the denominator in the metric label: “Started hires with known first source, 2026 Q3” is less viral than “Best source!” but vastly more useful.

The four attribution views worth keeping

Define which source view is authoritative for budgeting, which for funnel operations and which is context only. Do not let every analyst redefine “source of hire” in a separate dashboard. Maintain source history with event timestamps; preserve the raw record even when a normalized channel category changes.

Unknown is a category, not a cleaning error

Missing source metadata is nonrandom in many ATS setups: old records, referrals that enter via manual import and direct applications blocked by tracking permissions can behave differently. Replacing nulls with “direct” silently inflates one channel. Carry an unknown-source share next to every report and show how it varies by job, source type and time.

If 40 of 200 starts have unknown origin, the proportion is 20%. The identifiable 160 can tell you which source leads among known records. It cannot prove the same ranking holds across all 200. Run a sensitivity bound: if all 40 unknown cases were from the runner-up, would the ranking reverse? That gives leadership an immediate view of the attribution claim’s fragility.

UTM links tell you what was clicked, not what was persuasive

Tracking parameters are useful for paid-campaign hygiene. LinkedIn’s recruiting marketing guide recommends UTM-coded links and matching source information through the ATS to connect channel spend to applicants. But a clicked link records a touchpoint; it does not measure whether a friend, employer brand or earlier contact caused the action. LinkedIn channel tracking guide.

Maintain a source taxonomy with inbound, outbound, referral, agency, internal and unknown; retain campaign and exact landing URL separately. Document whether sourced prospects have the same application status as inbound applicants. Otherwise you compare differently constructed cohorts.

Privacy note: do not add unnecessary cross-site or invasive tracking merely to improve an analytics chart. Technical measurement must align with data protection and consent/cookie rules where applicable.

Worked example: the same eight hires change the “winner”

Illustrative cohort: eight started hires with known first and last touches. First-touch logs: five referrals and three job-board. Last-touch logs: two referrals and six job-board. A first-touch chart crowns referrals at 62.5%; a last-touch chart crowns job-board at 75%. Both percentages may be correctly computed and strategically misleading if presented as the single causal source.

Review a paired touchpoint trail for each hire to see who changed labels—not two unrelated aggregate totals. If a candidate was sourced two years ago but entered an unrelated requisition last month, explicitly choose the attribution window and rules for re-engagement. Make the decision repeatable rather than arguing about which team “deserves” credit.

A source report with an honest cost denominator

Spend divided by hires is meaningful only if spend and hires refer to the same channel definition, time window and eligible cohort. Agency fees paid on starts should not be compared with job board cash spend divided by applications. Internal recruiter time, creative production, CRM cost and referral bonuses are optional cost categories that must be either included consistently or reported separately.

Use leading evidence of candidate quality: qualified applications, interview completions and job-relevant outcomes—not volume only. Still avoid claiming the channel itself caused high job performance without a design that addresses selection and organizational confounders.

A monthly reconciliation that catches the quiet bugs

1. Export a fixed cohort of job applications and started hires; use stable IDs.
2. Preserve original source events with timestamps and raw labels.
3. Freeze a mapping from raw source to normalized channel for that report version.
4. Produce first-touch, application-trigger and candidate-reported views side by side.
5. Publish unknown-source share, duplicates, unlinked starts and unresolved imports.
6. Reconcile total starts to HRIS/headcount; investigate differences before comparing ROI.

For every change to the taxonomy or integration, run synthetic cases: referral then job board, two applications for one candidate, no tracking link, an agency duplicate, and an internal mobility hire. Unit test the facts before interpreting the pie chart.

The bottom-line conclusion that is easy to miss

If attribution improves after you change ATS configuration, your “channel efficiency” may change even when candidate behavior does not. That is a measurement change, not a recruiting miracle. Mark methodology breaks on charts and compare a bridged historical sample where practical.

Do not claim a proprietary causal breakthrough: the value here is a reproducible four-view reporting protocol and a sensitivity check for unknown records. Those procedures can make budget conversations more honest even before hiring analytics become statistically sophisticated.

Try the operational check

This small calculator runs entirely in the browser; no real applicant records are required. Use it to make a process assumption visible before reporting a conclusion.

Make this a repeatable standard, not a one-off hack

Capture the date, decision owner, source data, role family and exception reason every time you apply the method. Review the oldest unresolved examples with a second person. If the evidence points to another root cause, update the workflow rather than forcing a convenient metric to match your theory.

All cases and numerical values in this article are illustrative unless explicitly attributed. The operational checklists are a transparent practitioner synthesis of source guidance, not a claim of a universally validated test, legal rule or proprietary measured result. Jurisdiction-specific requirements require local review.

Research basis and further reading

Primary regulatory/professional guidance and explicitly identified vendor operational analysis. Sources are relevant reference points; the specific operational framework and synthetic cases above are our synthesis rather than findings directly measured by every linked source.

1. Ashby tracking an application source
2. Ashby source taxonomy
3. Ashby candidate versus application unit
4. Ashby source site attribution release
5. Ashby job dashboard source interpretation
6. LinkedIn UTM recruiting ad guidance
7. LinkedIn how job seekers discover roles
8. Indeed ATS attribution limitations
9. Ashby EMEA recruiter productivity source trends
10. ICO recruitment data processing

Apply the finding before the next decision

Choose a single open role, review real process events, and document one falsifiable improvement. Keep the decision in human hands and show candidates what changed in their experience—not just on a dashboard.