A confidential construction search fails through internal disclosure far more often than through recruiter error. Name three people, write down what each may say, keep interviews off the pattern, and decide in advance what happens when someone asks a direct question. Most leaks come from helpfulness, not malice.
A confidential construction search rarely fails where anyone expects. A contractor in Houston decides to replace a project executive who is running two live jobs and does not know. The search runs quietly for five weeks. Then the CFO, trying to be helpful, asks the controller to set up a background check budget code, and the controller mentions it to the office manager, who schedules travel.
Nobody did anything wrong. Within nine days the project executive knows, the two superintendents on their jobs know, and the owner on the larger job has asked a pointed question at a progress meeting. This is what a leak looks like, and it almost never starts with the recruiter.
What is a confidential construction search?
Confidential executive search for mandates that cannot be publicly posted. No advertisement, no job posting, no visible market activity. Advertising the role would itself cause the damage the search exists to avoid.
Three situations produce it. A seated incumbent who is being replaced and does not know. A bid or project milestone where signalling that preconstruction or operations is exposed would cost the firm work or credibility with an owner. And a market entry the firm has not announced, where naming the role names the strategy.
All three share one property. The cost of the information reaching the wrong person exceeds the cost of a slower, smaller pool. That trade is why the engagement exists. It is also why this is not a normal search conducted discreetly.
The mechanics differ throughout. Direct approach only. A shortlist selected, not filtered. Staged disclosure to candidates. And on the client side, a discretion plan agreed before the first call.
Who should know inside your own firm?
Three people or fewer, named in writing before the search begins, with a written note of what each one may say if asked. Every addition after that is a decision someone makes deliberately, not something that happens because a conversation made it convenient.
The usual three are the person the role reports to, whoever must approve the compensation, and one operational contact who can schedule without asking why. That third seat is the one firms most often fill badly, by using an executive assistant who supports several people including, sometimes, the incumbent.
Write down what each person may say. This sounds excessive until the moment someone is asked a direct question in a hallway, at which point an unprepared honest person will either lie badly or confirm by hesitating. Both are worse than a prepared, true, non-disclosive answer.
- The hiring manager, who owns the brief and the decision
- The compensation approver, usually a president or CFO, looped in once rather than continuously
- One operational contact who can schedule, and who does not support the incumbent
- Nobody else, until a named person decides otherwise and records why
The fourth line is the one that carries the weight. In practice, most confidential searches expand their circle by one person at a time, each time for a defensible reason, until eleven people know and no one made a decision to tell eleven people.
How does a confidential construction search leak?
A confidential construction search leaks through operational tells, not conversations. Calendar entries, a pattern of similar-length meetings, an unusual expense code, a badge visitor, or a hiring manager who suddenly cannot be found on Thursday afternoons. People in construction firms read these fluently.
Interview logistics are the largest single source. A seated project executive whose manager takes three ninety-minute meetings offsite in a fortnight, each with a person nobody in the office knows, does not need to be told anything. The team assembles the picture from fragments, and construction teams are unusually good at this because reading a site for signals is the job. Selah runs this as a chief estimator search.
The second source is systems. Background check requests, drug screen scheduling, offer letter templates, and travel bookings all touch people outside the named circle. Each is routine and each is a disclosure. Decide who processes these before the search starts, not when the offer is being prepared.
The third is the candidate side, and it is the one clients underestimate. A candidate told the client name too early, who then mentions the conversation to a colleague, can put the information into the same regional market the incumbent works in, whether that is Houston or Atlanta. Good practice is staged disclosure: sector, scale, delivery method and geography first; the name once the candidate is serious and has agreed to confidentiality.
| Leak channel | How it happens | The control |
|---|---|---|
| Interview logistics | A visible pattern of offsite meetings | Vary time, place and length; use neutral locations |
| Internal systems | Background checks, travel, expense codes | Name one processor inside the circle before starting |
| Circle creep | One more person, for a good reason, repeatedly | Written circle; additions require a recorded decision |
| Candidate side | Client name disclosed too early | Staged disclosure, name last, confidentiality agreed |
What happens when someone asks directly?
Decide the answer before it is asked. A prepared, honest, non-disclosive response holds. Improvisation does not, because a hesitation is read as confirmation and a denial that later proves false costs more than the original disclosure would have.
The workable form is true and bounded: we are always looking at how the team is structured, and I am not going to discuss individual roles. That is accurate in almost every firm, it does not deny anything, and it does not invite a follow-up.
What fails is the flat denial. A president who says there is no search running, and is later revealed to have been three weeks into one, has damaged something more durable than the search. The team that hears the denial and later learns the truth will apply that discount to everything else they are told, including the reassurance offered when the transition is finally announced.
It is worth being clear-eyed about the odds here. Not every confidential search stays confidential, and a plan is a probability improvement rather than a guarantee. What the plan does reliably is remove the avoidable failures, which are the majority.
What does a leak actually cost?
The incumbent’s tenure ends on their timetable, not yours. The seat opens into a hard market: 91.7 percent of the 1,041 contractors answering the salaried-hiring question in the 2025 AGC and NCCER workforce survey reported difficulty filling salaried positions.
That is the direct cost. It is the smaller one. The larger cost lands on the live projects. A project executive who learns mid-job that they are being replaced still controls information the firm needs and relationships the owner values.
A third cost outlasts both. Everyone now knows how a replacement is handled here. The strongest people in the building draw the obvious conclusion. The flight risk created by a badly handled exit often exceeds the problem it solved.
Who does not need this?
A firm replacing someone who has already resigned, or filling a genuinely new seat that can be announced, does not need any of it. Confidentiality has a cost in pool size and speed, and paying it when the role can simply be posted is a waste.
It also does not apply where the incumbent already knows and the transition is agreed. A planned succession, where the outgoing leader is part of the process, is a different and much easier engagement. Some of the best construction transitions run exactly that way, and firms that can do it should.
The discipline in this piece is for the narrow case where information itself is the risk. If that is where you are, the plan is worth an hour before the search starts rather than a crisis in week five. How a confidential search runs covers the recruiter side of the same work, and employers is the place to start the conversation.
Sources and further reading
- AGC, Associated General Contractors of America
- NCCER, construction education and research
- BLS wage data, construction managers (11-9021)
- BLS wage data, cost estimators (13-1051)