What is active sourcing, and when is it worth the effort?

What active sourcing actually is

Active sourcing inverts the direction of a hiring process. Instead of publishing a role and evaluating whoever responds, a sourcer defines the profile first and then goes looking for people who fit it — in professional networks, in communities and conference lists, in the company's own database of past applicants and silver-medalists, and through structured referral asks. The output is a list of named people and a reason each one is on it.

Two things follow from that inversion. The first is that the people on the list did not ask to be there, which changes the tone of every subsequent interaction and the amount of their time you can reasonably request. The second is that the quality of the list is entirely a function of the profile it was built from. Sourcing against a vague role produces a long list of plausible strangers, which feels like progress and is not.

It is worth separating sourcing from its adjacent activities. Employer branding changes who is willing to reply. Screening decides who continues. Assessment decides who is better. Sourcing does one job: it puts a specific person in front of a specific opening who would not otherwise have been there.

When it is worth the effort

Active sourcing is expensive in the one resource hiring teams have least of, which is attention. It earns that cost in a narrow set of conditions and burns it in the rest.

• The capability is genuinely scarce in the applying population — the people who can do the work are employed, not looking, and never see your posting.
• The qualified population is small enough to enumerate. If you can list the two hundred people in a region who have done this work, sourcing is a tractable search rather than a fishing trip.
• The search is confidential, or the role is new enough that nobody is searching for its title yet.
• You are deliberately widening beyond your usual channels, because the applicant flow you get reflects where you have historically posted rather than who can do the job.
• The cost of a slow hire is concentrated and known — a team blocked, a contract at risk — rather than diffuse.

When it is not

The commonest misapplication is sourcing into abundance. If a posting already produces two hundred applications and the team cannot get through them, adding sourced candidates adds work at the front of a queue whose problem is at the back. The bottleneck is evaluation capacity, and sourcing makes it worse while looking like initiative.

The second misapplication is subtler and more expensive: sourcing to compensate for a selection stage that cannot distinguish. When a team is unhappy with its hires and responds by widening the top of the funnel, the implicit theory is that the right person was never in the room. Sometimes that is true. Often the right person was in the room and the process could not see them, in which case a larger room produces the same outcome with more overhead.

The part teams underestimate: sourcing does not fix selection

This deserves stating flatly. The value of being selective depends on being able to select. A process that ranks candidates well converts a bigger pool into a better hire; a process that ranks them poorly converts a bigger pool into a more confident version of the same coin flip, because the extra candidates arrive as noise rather than signal. Volume multiplies whatever discrimination your selection stage already has, including none.

The selection research is a useful corrective here, in both directions. Meta-analytic work in personnel selection has spent decades estimating how strongly different assessment methods relate to later performance, and the 2022 re-analysis by Sackett and colleagues argued that long-accepted figures had been systematically overcorrected and the true relationships are more modest than the field assumed. The honest reading is not that assessment does not matter — it is that no method carries a hiring decision by itself, and the differences between methods are real but smaller than the marketing around them.

The practical test is one question, asked of your own process: when two candidates reach the final stage, can anyone state in writing which competency separated them and what evidence showed it? If the answer is a feeling, sourcing is not the intervention. If the answer is a rubric score against a task both people did, then more candidates is a genuine improvement, because each new one enters a machine that can place them.

The handoff, which is where the value actually is

Sourcing and assessment are usually owned by different people, and the seam between them is where sourced candidates are lost. A sourced person has agreed to a conversation, not to a process. Handing them an unexplained sixty-minute test as step two is the fastest way to convert a warm reply into silence.

A handoff that survives contact needs four things, and none of them is technology. Both sides must be working from the same role definition, so the profile the sourcer searched against is the profile the task measures. The task must be short enough to be a fair ask of someone who did not apply, and recognizable enough as real work that doing it feels like a preview rather than a hazing. The invitation must say why this person specifically, what the task involves, roughly how long it takes, and what happens to what they produce. And the result must come back fast, because a sourced candidate's alternative is doing nothing, which costs them nothing.

Where accommodations are needed, they belong in the same message rather than in a form nobody reads. Extended time and format alternatives are cheaper to grant early than to retrofit after someone has already dropped out.

One standard, two intakes

A sourced candidate and an applicant should meet the same rubric. This sounds obvious and is routinely violated, usually with good intentions: the sourced person was hard to find and came recommended, so the task gets waived, or scored generously, or replaced with a conversation. At that point the process has two standards, and the comparison the whole apparatus exists to produce no longer holds.

There is a record-keeping dimension too, which teams hiring in the United States meet sooner than they expect. Federal contractor rules define an "Internet Applicant" narrowly — an individual who expressed interest electronically, whom the contractor considered for a particular position, whose expression of interest indicates they meet the basic qualifications, and who did not withdraw before an offer — and the regulation is explicit that this covers expressions of interest however they are made, if the contractor considers electronic ones for that position. Sourced people can fall inside that definition. Separately, the Uniform Guidelines' four-fifths convention treats a selection rate for a group below four-fifths of the highest group's rate as evidence of adverse impact, and sourcing shapes the pool those rates are computed over.

Obligations vary by jurisdiction, this is not legal advice, and SkillCort makes no claim that using it puts an organization in compliance with anything. The point is narrower: sourcing decisions are part of the selection record whether or not anyone is keeping one.

What we do and do not do here

SkillCort is not a sourcing tool. We do not search networks, we do not build candidate lists, we hold no cross-client candidate reputation of any kind, and nothing we do makes a stranger more likely to answer your message. If sourcing is the problem you are solving this quarter, the tools for it are elsewhere and we will say so.

What we own starts one step later, at the point where you have a person and a question about them. A role blueprint that names the competencies and their weights gives your sourcing a profile worth searching against. A work-sample task built from that blueprint gives a sourced candidate something worth an hour. A rubric and calibrated evaluators mean the answer does not depend on who happened to read it, and the evidence file means the reason survives the week the decision was made.

Put plainly: sourcing determines who is in the room, and we have no view on that. Whether the room can tell them apart is the part we work on.

Sources & further reading

• 41 CFR 60-1.3 — Definitions, including "Internet Applicant" (OFCCP) — U.S. Code of Federal Regulations
• 29 CFR 1607.4 — Information on impact (Uniform Guidelines, four-fifths rule) — U.S. Code of Federal Regulations
• Sackett, Zhang, Berry & Lievens (2022), Revisiting Meta-Analytic Estimates of Validity in Personnel Selection — Journal of Applied Psychology