How Executive Search Researchers Actually Find Candidates

The candidates you reject are visible. You can look at the list, second-guess a cut, ask for the reasoning. The candidates you never saw are invisible, and they are the expensive ones — a search fails on the person who was never in the pool far more often than on the person who was in the pool and screened out.

That makes candidate sourcing a recall problem before it is a filtering problem. And recall is mostly decided early, by how wide the net was thrown before anyone started judging.

The discovery problem

Before a single person is evaluated, someone has to assemble the pool. A well-run first pass is wide — wide enough that a plausible candidate is not missed because the initial framing was too narrow. How wide varies with how well-mapped the sector is; a mature vertical with visible org structures yields more than a fragmented one.

That can look excessive for a search that will deliver six names. It is not generous; it is the cost of the funnel. A wide pass that is mostly wrong is recoverable. A narrow pass that missed the right person is not, because nothing downstream goes looking for people who were never entered.

Note what the wide pass is not: a search-and-judge loop. Discovery and evaluation are different activities, and running them together is how pools end up narrow — assess each name as you find it and you calibrate, within the hour, to the type of person you have already been finding.

The source layers

No single source produces a pool. Each has a characteristic blind spot, and the point of working several is that their blind spots are different. The grouping below is how we think about it rather than an industry taxonomy — the layers matter more than the labels.

Professional profiles. The obvious starting layer and the one everyone uses, which is exactly why it cannot be the only one. It is self-reported, unevenly maintained, and biased toward people who are professionally visible. That correlates with job-seeking and with sector, not with capability.

Company org structures. Working from the company inward rather than the person outward. Who runs operations at each of the forty companies that plausibly produce this profile? Slower than the first layer, and it reaches the people who are not looking — often where laterals worth a second look come from.

Announcements and appointments. Press releases, trade publications, and internal-promotion announcements are the layer that tells you what someone actually took on and when. They are also the layer that produces citable evidence rather than self-description.

Filings and public records. Board appointments, regulatory filings, and disclosure documents. Narrow in coverage, high in reliability. When a claim needs to hold up in front of a client, this is where you want it to come from.

Published work and speaking. Conference programs, technical papers, patents, industry panels. This layer is uneven — plenty of strong operators never publish anything — but when it hits, it tells you how someone thinks about the problem, which no title conveys.

Adjacent-industry mapping. Deliberately looking one sector over, where the operating constraints are similar even though the label is different — medical devices and aerospace share a regulatory posture. In our experience this is a question that has to be asked on purpose; it rarely gets answered by accident.

Why filtering early is the expensive mistake

Under deadline pressure, narrowing early feels like progress. Every name removed is a name you do not have to research.

The problem is asymmetry. A person wrongly kept costs a few minutes at the next stage, where they get removed anyway. A person wrongly cut costs the search, silently, and no stage revisits the excluded.

So the sequence is: assemble first, judge second. Staging the filter properly is what makes the wide pass manageable rather than overwhelming.

The expensive mistake is invisible by construction: you can audit who you rejected, but never who you failed to consider.

First-pass filter versus deep research

These are different questions asked of different people, and conflating them is what makes discovery feel unmanageable.

The first-pass filter asks one thing: could this person plausibly qualify? Function, seniority band, sector, geography. Facts you can establish in a minute without weighing anything. Most of the pool leaves here, and it should — you are removing people no amount of research would rescue.

The deep pass asks what someone has actually done, evaluated against each criterion in the brief individually, with a source attached to every confirmed claim. Scope rather than title. Evidence rather than description. This is the expensive work, and it runs on the survivors precisely so it can be done properly rather than skimmed across the whole pool. What that screening pass involves is a discipline in its own right.

Where candidates get missed

Single-source dependence. If the pool came from one platform, it inherits that platform’s population. Everyone who maintains a sparse profile is absent, and you have no way of knowing who.

Title matching. Titles do not mean the same thing across company sizes or sectors. Searching for “VP Operations” finds people labeled VP Operations. It misses the plant director running three sites and the general manager whose scope is larger than the title suggests. Search on scope; treat titles as a hint.

Entity resolution errors. Two people with the same name in the same industry, and a research pass that merges their histories. The result is a candidate who looks remarkable and does not exist. This is the failure our own QA is built hardest against, because it produces a confident, detailed, wrong entry, and it passes every downstream check that is not specifically looking for it. It is also why what counts as a confirmed claim has to be defined before the research starts, not after.

Geographic and language narrowing. A pool assembled in one language misses operators whose entire public footprint is in another. In cross-border searches this quietly removes whole markets.

Stopping at the first plausible pool. Discovery ends when the search feels sufficient, and “sufficient” arrives much earlier than it should, right around the point where the pattern starts to feel familiar. That feeling is not coverage. It is fatigue.

What good discovery looks like

Good discovery is boring and systematic: several source layers run to their end rather than to the point of comfort, scope-based rather than title-based matching, and evaluation held back until the pool is assembled.

And afterwards, one uncomfortable question, asked honestly: if the right candidate for this role is not in this pool, where would they have been?

If that question has an answer that was not searched, discovery is not finished. This is the part of the process that tends to get compressed first when research capacity is thin — and it is the part where compression is least visible and most expensive.

Related reading: where discovery sits in the wider practice, where a stale profile breaks recall, hiring operations executives, what candidate sourcing services actually include, headhunter versus recruiter, how a CEO search actually works, what a good executive search shortlist contains, how AI is actually used in executive candidate research, what candidate sourcing tools are actually built for, and what active sourcing actually means in executive recruitment.