Pull the source-of-hire report out of your applicant tracking system (ATS). The careers page can come out of it as your best recruiting channel. The employee who forwarded the job link and the recruiter whose message started the conversation both lose the credit, because the candidate finished on your site and the ATS may have stored only that last step.

Read that way, the report sends budget toward the careers page and away from the people who found your hires. I’d keep two things on every candidate: the route they applied through, and the source that first brought them to your team. Only the second one is source of hire. With that in place, the reading that pays off is where each source’s candidates stop in your process. That stop is something your team can change, and I’d change it before cutting the source.

What counts as the source of a hire?

Source of hire is the channel that first brought a person you hired to your team. The channel they applied through is often a different one, and the application route is what an ATS usually fills in when someone submits a form.

A candidate can hear about a role from a friend who works for you and then apply on your careers page. The careers page is the route they took, and the referral is the origin that earned the hire.

So I’d keep two fields apart on every candidate record. Origin is the channel through which your team first heard of the candidate, and it’s the one a source-of-hire report should count. Application route is where they submitted, which the ATS usually records on its own and which is still worth keeping for the careers page team.

The origin field needs a fixed list of values, with one value for each thing your team spends money or recruiter hours on. That usually means splitting entries a lot of teams lump together:

  • LinkedIn job posts and LinkedIn messages from a recruiter, which cost different things and reach different people.
  • Each agency by name, because each one brings its own candidates and its own stage counts.
  • Internal moves, as a value of their own.
  • Past applicants, who need a rule more than a value: a person your team found again in the ATS keeps the origin on their first record, and a second field notes what brought them back this time.

Leave “Other” off the list. Every candidate filed under it is a source you will never be able to read back.

Record the source where the candidate enters.

The origin is easiest to get right at the moment a candidate enters your pipeline, because the person adding them already knows how they came in. So the person who adds the candidate sets it. For each way in, this is who that is, and when:

How the candidate came in Who records the origin When it gets recorded
Found by a recruiter or sourcer The person who found them When the candidate is added to the role
Referred by an employee The employee, through the referral form When the referral is submitted
Put forward by an agency The recruiter who accepts the submission, naming the agency The day the submission arrives
Applied on their own The candidate on an optional form question, confirmed by the recruiter When they apply, and again at the first call
A past applicant reopened for a new role The recruiter who reopened the record, in the field for what brought them back When the record is added to the new role
An employee moving internally The recruiter on the role When the internal application comes in

People who applied on their own are the one case where the first value needs a second pair of eyes. A candidate answering a form question may pick whichever option roughly fits, so I’d treat the answer as a draft: the recruiter confirms or corrects it on the first call.

Read each source stage by stage.

A source-of-hire count says where your hires came from. It can’t say whether an agency’s candidates fell away at the hiring manager interview or at the offer. The stage where they fell away is the part you can change.

So for each origin, build a row of stage counts from every candidate who entered in the same window, and write down where its candidates stop:

  • How many candidates entered from that origin.
  • How many passed each of your stages, from the recruiter screen through to an accepted offer.
  • How long they spent between stages.
  • What the origin cost over the same window.

Then compare the shape of each origin’s row with the others. A job board that sends many applicants and loses almost all of them at the screen has a different problem from an agency whose candidates clear the screen and then stop at the hiring manager’s interview, even if both brought in one hire last quarter.

Repair the stage where each source loses candidates.

A source whose candidates keep stopping at one stage usually points at a problem in that stage. Before I’d move money away from it, I’d change the step where its candidates fall away and give the source one more round of roles to show whether the change worked.

These are the stalls I’d look for first, and the step each one usually points at:

  • Job board candidates who drop at the recruiter screen usually point at the post. If it describes a broader role than the one you’re hiring for, rewrite it before you blame the board.
  • Agency candidates who clear the screen and stop at the hiring manager usually point at the brief. Invite the hiring manager to the agency’s briefing call for the next role, and send the agency the reason for every rejection.
  • Referred candidates who wait days for a first call point at how your team answers referrals. Book that call in the week the name arrives, because the employee who sent it is waiting to hear too.
  • Sourced candidates who take a first call and then withdraw usually point at the gap between the first message and the role the call described.

Cut the source only after a fix has failed, and move its money and hours to the origin whose candidates get furthest for the time your team puts in.

When the stage counts leave you with more channels than your recruiters can look after, a separate guide on pruning the channel list works through how many to run and which to drop.

Where Metaview fits in source of hire.

For a stall at an interview stage, the reason is often in what was said there. An ATS export holds the count of who stopped and none of the conversation.

Metaview Reports can answer questions about the interviews your team has on record, and a question can name the candidates it is about. That lets you read what the agency candidates on your export said in the interview where they stopped.

Metaview captures every spoken word. With that record, a hiring manager’s view that an agency’s candidates were thin on one must-have can be held against the candidates’ answers, and their answers show whether to rework the agency’s brief or look for a different agency.

Which conversations get recorded is up to the people on the call, under the consent rules your team follows and local law. On a video interview that is being recorded, the Metaview Notetaker comes in as a participant the candidate and interviewers can all see, and it leaves a transcript with structured notes once the call ends.

Stage counts are the data a talent leader usually does get. Metaview’s chief executive put the other side of it this way:

Interviews are vital when building an exceptional team, and being an effective interviewer is a highly-valued skill. But there is so little data available to leaders about these crucial moments.”
Siadhal Magos Siadhal Magos Co-founder and Chief Executive Officer · Metaview

I’d bring that record into the next agency review beside the stage counts, and go through the missed must-have with the agency one candidate at a time.

If your team finds candidates with Metaview AI Sourcing, the person who adds one of them to a role sets the origin, and the past-applicant rule still holds. Someone new to your team gets AI Sourcing as their origin. Someone your ATS already held keeps their first origin, with AI Sourcing noted as what brought them back.

Both cases come up, because the sourcing agent works from one natural-language prompt and searches the web and your connected ATS, and also earlier Metaview conversations from your own workspace. It returns each candidate with its reasoning. Who on that list gets a message, and who suits the role, is for the recruiter to decide.

See the candidates your source field will be recording.
A demo puts one natural-language prompt to the sourcing agent and walks through the candidates it returns and the reasoning on each one.
See it live

Audit a quarter of origins against their records.

Each quarter, pull twenty candidates who entered in that window, including people you didn’t hire, because the stage counts run on every one of them. Open each record and set the origin field beside the evidence it holds: the referral submission, the agency email, or the sourcer’s first message.

Correct every origin that disagrees with its evidence, and note who set it. Then write the number of corrections at the top of that quarter’s stage counts, where anyone reading them sees it first.

A quarter with a correction or two is one you can fund and cut from. A quarter with many tells you to fix the point where origins get set before any budget moves, because until then the report is still crediting the careers page with hires a referral or a recruiter started.

See it in action.

Read what one source’s candidates said at the stage they stopped.

A demo walks through Metaview Reports answering questions about candidates you name, from their recorded interviews, which your stage counts can’t show.

Frequently asked.

Who gets the credit when an agency submits a candidate already in your ATS?

The candidate keeps the origin on their first record, and the agency goes in the field for what brought them back. Whether a fee is owed is a question for your agreement with the agency, and the source report shouldn’t be what settles it.

Should internal hires count in source-of-hire data?

Record them, with internal move as their origin, and leave them out when you compare outside channels. An internal candidate skips most of the stages an outside one goes through, so putting the two side by side makes every outside channel look slow.

Who should own the source-of-hire field on a recruiting team?

One person owns the list of origin values, usually whoever runs recruiting operations, and new values go through them. Each recruiter owns the entries on their own candidates. A list anyone can edit soon holds two spellings of every agency, and the report splits each one in half.

Can source-of-hire data be used to compare recruiters?

Only within one role family. A recruiter filling senior engineering roles leans on sourcing and agencies, while one filling support roles leans on job boards, so their mix of origins says more about their roles than their work. Compare how each recruiter’s candidates from the same origin move through the same stages.

How often should you review source-of-hire data?

Read the stage counts once enough candidates from each origin have finished their process: quarterly works for roles you hire all the time, and twice a year suits roles you fill rarely.