The Research Bottleneck in Boutique Executive Search
Ask a managing partner at a boutique search firm what limits growth and the answer is rarely business development. Most boutiques could take more mandates than they currently run.
What they cannot do is research them all properly at the same time.
Research is the longest and least compressible stage of a search, and at a firm of eight to twenty people there is no bench. So research capacity sets the ceiling on concurrent searches. That makes it a business constraint wearing the costume of an operational one.
Who actually does the research
At a large firm, research is a function: dedicated researchers, a knowledge management layer, sometimes an offshore support team. At a boutique it tends to be one of three arrangements in practice, each with its own problem.
A dedicated researcher, or one and a half. The cleanest version, and it is a single point of capacity. When they are loaded, the firm is loaded. When they are out, so is the firm’s throughput. Any search that starts while they are saturated gets research that was fitted around other work.
Associates who research between other duties. Research competes with client coordination, scheduling, and candidate management. All of those have visible deadlines. Research does not. It has a soft deadline and an invisible quality floor, which means it is always the thing that gets compressed when the week goes wrong.
The partner. Common at the smallest firms and the most expensive arrangement in the building. A partner doing market mapping is doing work at a fraction of their marginal value, and the time is coming out of business development and client relationships — the two activities that actually determine whether the firm grows.
The time it takes
Ask a managing partner how long mapping, discovery, first-pass filtering and deep qualification take on one search, and the answer comes back in days rather than hours — how many depends on how well-mapped the sector is and how specific the brief was.
You know your own number better than any industry figure would tell you, so run the arithmetic with it. Take the days your firm actually spends, multiply by a fully-loaded day rate for whoever is doing the work, and the research stages of a single search land somewhere in the four figures. Partner time doing the same work is a multiple of that, before counting what the partner was not doing instead.
What makes this hard to manage is that the cost is real and mostly invisible. It does not appear as a line item. It appears as a firm that runs six searches when it could run nine.
Quality that varies by week
The cost that does more damage is not the hours. It is the variance.
The same firm, the same researcher, the same standards — and a search that starts in a quiet week gets a materially different research pass than one that starts in a week with two client presentations and a candidate withdrawal.
The compression is predictable in where it lands, because the stages are not equally visible:
- The discovery pass narrows. A shorter list instead of the full one. Nobody notices, because the missing people are not on any list. This is the most consequential compression and the least detectable one.
- Sourcing gets looser. Claims that would have been checked against a second source get accepted from one. The entry reads identically either way.
- The lateral pass gets skipped. It is the easiest stage to drop because nothing breaks visibly when you do. The client receives six on-spec candidates and never learns about the two laterals that were in the near-miss pile.
- QA collapses into a proofread. Formatting gets checked; attribution does not. Mis-attributed claims survive to delivery.
None of that shows up as a failed search. It shows up as a search that took longer, or a shortlist the client was lukewarm about, or a role that got filled by someone the client found themselves.
Quality that depends on who had bandwidth that week is not a standard. It is a coin flip you are not watching.
Why it is structural, not a staffing problem
The obvious fix is to hire another researcher. Sometimes that is right. Usually it moves the constraint rather than removing it, for reasons that are structural to how boutiques work.
Demand is lumpy and headcount is not. Mandates arrive in clusters. Hiring for the peak means carrying idle cost in the troughs; hiring for the average means the peaks compress quality. There is no headcount number that solves both.
The skill takes time to build. A researcher who can reconstruct operating scope from filings and announcements, recognize a merged identity, and tell a confirmed claim from an asserted one is not a first-month hire. The ramp is months, which means hiring responds to demand roughly one cycle too late.
Quality does not survive the handoff. Splitting research across more people without a written standard produces more output at wider variance, which is the problem rather than the solution.
So the constraint holds. Firms respond by declining mandates, running searches slower than they would like, or accepting variance and hoping it lands in the right places.
What actually changes it
Only two things move a structural constraint: raising the ceiling, or removing work from underneath it.
Raising the ceiling means capacity that is not headcount-shaped — able to absorb a peak without carrying idle cost through the trough.
Removing work means separating the parts of research that are mechanical from the parts that require judgment. Assembling the pool, checking scope against sources, evaluating candidates one criterion at a time, tracing every claim to a citation — this is disciplined, teachable, and structurally repeatable work. Deciding who is right for the role is not, and never will be.
That split is why the tooling question has become worth taking seriously, and also why most of the tooling on offer misses. The value is not in a system that produces candidate names quickly. It is in one that does the mechanical stages to a standard that does not vary with the week — and hands back a sourced, honestly-flagged picture for the judgment that was always yours.
It is also the design constraint behind how Cerna works: a fixed per-search cost for the research stages, which is a different shape of commitment from a headcount decision. What that costs is deliberately set against the week of research time it replaces, not against a subscription.
Related reading: what one metro’s firms look like up close, what the constraint looks like from year one, what a boutique firm can actually compete on, how to tell search firms apart, how executive search researchers find candidates, retained versus contingency search, which sets how much research a given mandate can fund, what candidate sourcing tools are actually built for, and what CFOs earn, and what actually sets the number.
Frequently asked questions
Why does research capacity become the bottleneck at boutique search firms?
Because it's the stage of a search that takes the longest to do properly and shrinks the least under pressure, and a firm of eight to twenty people has nobody in reserve to absorb a surge. Whoever can research gets fully booked, and that ceiling — not partner bandwidth, not client demand — ends up deciding how many searches the firm can actually run at once. It never shows up as a cost on a P&L; it shows up as mandates the firm quietly turned down.
Who typically does research at a boutique executive search firm?
One of three arrangements, each with its own problem: a dedicated researcher (or one and a half), which is a single point of capacity; associates researching between other duties, where research gets compressed because it has no visible deadline; or the partner, the most expensive arrangement, doing work at a fraction of their marginal value at the direct cost of business development.
How does a compressed research pass actually show up in a search?
In four predictable ways: the discovery pass narrows to a shorter list nobody notices is short, because the missing people aren't on any list; sourcing gets looser, with claims accepted from one source instead of checked against a second; the lateral pass gets skipped since nothing breaks visibly when it is; and QA collapses into a proofread, checking formatting rather than attribution.
Why doesn't simply hiring another researcher fully solve the bottleneck?
Demand is lumpy and headcount is not, so there's no staffing level that solves both the peaks and the troughs. The skill also takes months to build, meaning hiring responds to demand roughly one cycle too late. And splitting research across more people without a written standard produces more output at wider variance, which is the problem rather than the solution.
Which part of a compressed research pass is hardest to fix later?
A narrowed discovery pass. A loose source can be re-checked, a skipped lateral pass can be run late, and a thin QA pass can be redone — but a candidate who was never entered into the pool in the first place is never missed. No downstream stage goes looking for them, and no review surfaces them.