A construction interview panel tests only what its members are assigned to test. Five people asking the same three questions is one interview held five times. Give each interviewer one evidence area (commercial, field, people, owner-facing), cap the loop at three rounds, and name a single decision owner before the first conversation.
Round five, late on a Thursday. The chief financial officer leans back and asks the candidate to describe a difficult owner. The candidate has now answered that question four times, and the fourth telling was better than the first, because practice helps. Nobody on your interview panel has asked about a pay application, a buyout or a lost sub, and nobody is going to. Each of them assumed somebody else had.
Why does a large interview panel test less than a small one?
Nobody on it owns a subject. Each interviewer falls back on the questions any interviewer can ask (background, strengths, the difficult owner), so the same thin layer gets tested five times and everything underneath it goes untested.
The usual defense of a wide loop is rigor: more eyes, more chances to catch a problem. In practice the eyes all look at the same place. A president, an operations vice president, a chief estimator, a controller and a senior superintendent each bring a different expertise into the room and then set it aside to ask about leadership style. Leadership style is what a personality assessment already describes, and its report gives the panel questions, not a verdict.
The result feels thorough. Five people met the candidate, five people liked the candidate, and the debrief is a round of impressions that happen to agree. Agreement among people who gathered the same evidence is one data point, counted five times.
The gap is between functions, not within them. The controller knows what a clean cost report looks like and the superintendent knows what a project manager who has never run a pour looks like. Neither was asked to find out. A finalist’s 90-day plan falls into the same gap: with no area assigned, the panel grades its polish instead of its judgment.
Who should sit on a construction interview panel?
Four people at most, each assigned one evidence area before the candidate arrives: commercial, field, people and owner-facing. The assignment matters more than the seniority of whoever holds it. An interviewer without an area will ask about the difficult owner.
| Interviewer | Area owned | What they report |
|---|---|---|
| Controller or CFO | Commercial: forecasts, billing, change-order paper | Whether the candidate’s numbers held up when pressed |
| General superintendent | Field: sequencing, safety, how they treat a foreman | Whether the field would follow this person |
| Operations VP | People: who they hired, developed and let go | Names, dates and what happened to each |
| President or project executive | Owner-facing: bad news, negotiation, repeat work | How they would sound in front of your owner |
The commercial seat is the one most often filled by the wrong person. The controller, or whoever signs the work-in-progress schedule, is the member of your firm trained to hear a soft forecast, the discipline that bodies such as CFMA exist to develop. Put that person in the room with a cost report and a candidate, and you will learn more in forty minutes than in two rounds of general conversation.
Each interviewer needs three things in writing:
- The one area they own, and the instruction to stay inside it.
- Two or three questions tied to a job the candidate ran. The project manager questions are built that way.
- A line on the shared interview scorecard, filled in before they talk to anyone else.
Scoring before the debrief is the part people skip. Once the president has said “I liked that one,” the superintendent’s doubt about sequencing becomes a harder thing to say out loud.
How many rounds should the loop run?
Three, with two of them on the same day where possible. In a worked example, a five-round loop scheduled one conversation per week takes 35 calendar days from first interview to last. A three-round loop with two rounds held together takes 12.
The arithmetic is plain and built for illustration. Senior people are hard to schedule. One is on a job walk, one is in a bid, one is traveling. Take a contractor with its office in Houston and its largest project in Charlotte: every round that needs both the president and the operations lead is a flight or a week’s wait, so five rounds at one a week is 35 days. Put the commercial and field interviews on one afternoon, hold the people conversation by video within the week, finish with the decision owner, and the same evidence arrives in 12.
Those 23 days are not free. In the 2025 AGC and NCCER workforce survey, 91.7 percent of the 1,041 contractors answering the salaried-hiring question reported difficulty filling salaried positions. The Bureau of Labor Statistics tracks construction job openings every month. A candidate good enough to reach your round four is in someone else’s round two, which is why interview speed decides searches that shortlist quality should have decided.
What does a wide panel cost a confidential search?
Every added interviewer is one more person who knows the seat is open, and on a replacement search the seat is occupied. A five-person loop means five calendars, five assistants and five chances for a careless remark in a job trailer.
On an ordinary hire the cost of a wide loop is time. On a replacement it is exposure. The incumbent still runs work. The owner on that work has not been told. A senior superintendent who has spent two hours with a candidate for the boss’s job knows something that cannot be un-known, and that superintendent has a crew, a spouse and a favorite subcontractor.
The discipline for what your team learns during a confidential search applies with extra force to the panel: name the people, tell each what they may say, and hold the conversations off site and off the shared calendar. The four evidence areas still need covering. They do not need four insiders. A trusted outside advisor can take the commercial conversation, and the decision owner can take two areas personally.
One decision owner, named before the first interview
Somebody has to be able to say yes. In most firms that is the executive the seat reports to, and the name should be written down before the candidate’s first conversation, along with a plain statement that the other interviewers report evidence and do not vote.
Panels drift toward consensus because consensus feels safe. It also hands a veto to whoever is least comfortable, and the least comfortable person is often the one whose territory the new hire will change. A decision owner hears that discomfort as information about the firm, weighs it, and still decides. Keep the scored evidence afterward, because it is the first thing to reread if the result is a failing hire at 90 days.
Write the panel on one page before you open the search: four areas, the name against each, the order of rounds, and the owner of the decision. If the seat is one you cannot advertise, it needs confidential executive search for mandates that cannot be publicly posted, so bring the page to the first employer conversation.
If nobody on the panel was assigned anything, what you held was a reception.
Sources and further reading
- CFMA, Construction Financial Management Association
- AGC survey, 2025 workforce survey with NCCER, national results
- BLS JOLTS, job openings and labor turnover data