A construction project manager’s span of control is set by contract value, delivery method and stage, not by a count of projects. Most contractors overload the seat when backlog grows faster than the bench, and the first casualty is the forecast. Plan the project manager count from the backlog eighteen months out, not from the org chart today.
The question that decides a construction project manager’s span of control is almost never asked in the form it matters.
Contractors ask how many projects a project manager can run. The honest answer is that nobody knows, and the reason nobody knows is that the person carrying too much does not report it. They report a forecast that has not changed since last month, on a job where a great deal has changed, and the fact that the number went quiet is read as good news.
The surety reads it differently. So does the construction lender, and the CPA doing the year-end WIP review. They are looking at the same quiet number and asking a different question: how many jobs is that person actually on, and who is watching the ones they are not?
Why is there no standard number of projects per project manager?
Because a project is not a unit of work. A construction project manager’s span of control is bounded by contract value, delivery method, stage and distance between sites, and one number cannot hold all four. One $60 million negotiated healthcare job is a full load. So are three $6 million tenant improvements.
Take the four variables in turn and the reason becomes clear.
Contract value sets the volume of decisions per month: buyout packages, change orders, pay applications, subcontractor coordination. A $60 million job at peak generates roughly ten times the commercial traffic of a $6 million one, but not ten times the meetings, which is why value alone misleads.
Delivery method sets who owns the risk. A guaranteed maximum price job carries the contractor’s own contingency and a client who expects to see it managed; a hard-bid lump sum job has already sold the risk and now has to hold it. The FHWA construction program guidance for federal-aid work sets out how much of that administration falls on the contractor’s project staff, and it is not small.
Stage is the variable most planning ignores. Preconstruction and buyout are decision-dense; the middle of a well-bought job is comparatively quiet; closeout is dense again and unpaid. A project manager with one job in buyout and one in closeout is carrying two peaks at once, and the count says two.
Distance is the honest constraint. A project manager who has to drive ninety minutes between sites in the Dallas to Fort Worth corridor or across metro Atlanta loses a working day a week to the road, and the jobs on the far end of that drive get the weekly visit rather than the daily one.
None of which prevents the number being set anyway, usually by whatever backlog arrived that quarter.
Where does the span of control actually break?
At the second concurrent peak. A project manager can usually absorb one job entering buyout while another sits mid-stream. The seat breaks when two jobs peak together, or when a third is added because the backlog said so and the bench did not. The break is silent, and it reaches the forecast before the site.
The mechanism is worth stating plainly, because it explains why the problem is invisible from the office.
A stretched project manager does not stop working. They triage. The site gets attention, because the site is loud, and the monthly cost-to-complete gets built from what was written down last month, because that is the only version there is time for. The site team, asked what is still outstanding that has to go into the forecast, often cannot say yet: the thing that will cost the money has not announced itself. So the forecast gets built from the paper, and the part that would have changed it stays on site.
Under percentage-of-completion accounting, that forecast is the firm’s reported profit. The CFMA treats forecast integrity as a governance control precisely because a project manager under load will, without any dishonesty, produce a number that is stale rather than wrong, and a stale number reads as stable.
A surety underwriter reviewing the WIP schedule sees a job whose percent complete has advanced and whose estimated margin has not moved in four months. That is not reassuring. It is the signature of a seat nobody is fully in.
| Signal on the WIP schedule | What it usually means in the seat |
|---|---|
| Margin flat for three or more months on an active job | Forecast rolled forward, not rebuilt |
| Costs incurred running ahead of billings | Pay application slipping; the paperwork is behind the work |
| Change-order log with nothing pending | Changes happening on site, not yet priced |
| Two jobs on the same manager both in the last 15 percent | Two closeouts at once; retention is at risk on both |
| Underbillings rising on the newest job | Buyout incomplete; the budget is still the estimate |
Every row in that table is something a gatekeeper sees before the contractor’s own leadership does, because the gatekeeper is reading the schedule cold. The contractor is reading it with the assumption that the project manager has it under control. The seat that is supposed to see across all of them at once is the construction project director, and a firm without one is relying on the assumption alone.
The assumption is usually right, until the month it is not.
How should a contractor plan the project manager count?
From the backlog, eighteen months out. Map every awarded and probable job by expected stage per quarter, count the peaks that coincide, and staff to the peaks rather than the average. A contractor that plans the bench from today’s org chart is always one project manager short in the quarter the backlog converts.
The arithmetic is simple and almost nobody does it. Selah runs confidential searches for a senior project manager.
Start with the awarded backlog and the probable pursuits, weighted by hit rate. For each job, mark the quarters in which it will be in buyout, in peak construction and in closeout. Stack the quarters. Wherever two decision-dense stages coincide on the same manager, that manager is overloaded in that quarter whatever the project count says. The AGC hiring outlook records that contractors expect to add staff, and the Census Bureau spending series shows where the volume is arriving, but neither tells a given firm which quarter its own bench breaks.
- Count peaks, not projects. A quarter with two buyouts on one manager is the constraint, not the annual total.
- Price the distance. A job ninety minutes from the office is a different load from the same job across the street.
- Treat closeout as a peak. It is unpaid, it is where retention and warranty exposure sit, and it is where stretched managers stop showing up.
- Plan the hire from the backlog conversion date, not from the day the overload becomes visible. A construction executive search takes weeks the overloaded quarter does not have.
Consider a regional general contractor in Nashville carrying $120 million of annual volume across four project managers. The average is $30 million each, which sounds comfortable. Stack the stages and the picture changes: in the second quarter of next year, two of the four are each carrying a new job entering buyout on top of a job in closeout. Those two managers are each carrying roughly $60 million for that quarter, double their sustainable load, and the other two have capacity they cannot lend because their jobs are ninety minutes away.
The averaged view says four project managers is correct. The stacked view says the firm needs a fifth for one quarter, or needs to sequence its pursuits so the peaks do not coincide, and needs to decide which before the second job is awarded.
That decision is a hiring decision, and it has a lead time.
What does the overload look like from outside?
Like a firm that has outgrown its bench. A surety, a lender or an owner’s representative reads project managers stretched across concurrent peaks as a capacity problem, and capacity is what a bond program prices. Backlog added faster than project management depth is, in the underwriter’s terms, exposure without the ability to manage it.
The general shortage compresses the fix. 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, and the AGC release records project managers among the hardest seats to fill. The BLS occupational profile puts the construction manager pool at a size the whole industry is drawing on at once.
So the overloaded quarter cannot be fixed in the overloaded quarter. The search has to start when the backlog converts, which is usually before the overload is visible to anyone inside the firm. How bonding capacity and bench depth interact is worked through separately; the short version is that the surety notices the missing project manager before the contractor does, and the prequalification leader is the seat that has to explain it.
There is also a version of this problem where the seat cannot be advertised. A contractor that has decided one of its four managers is the one whose forecasts have gone quiet, and needs to replace them while they are still running a live job, cannot post the role without telling the person and the client. Selah Talent Partners runs those as confidential executive search for mandates that cannot be publicly posted, and the confidential construction search piece sets out how.
Selah works with contractors and construction consultancies across the United States on preconstruction and estimating, project and construction management, and cost and commercial management. Candidates are never charged a fee at any stage.
This is not a piece for a specialty contractor running many small jobs through superintendents with a single project manager coordinating billing. That model has its own arithmetic, and the span is set by the superintendent bench rather than the project manager one. The analysis here applies to general contractors and construction managers where the project manager carries the commercial position on each job.
Count the peaks
The org chart says four project managers. The backlog says that for one quarter next year, the firm has two.
That gap is where fades go unreported, where the WIP schedule goes quiet, and where a surety starts asking questions the president has not yet thought to ask. The fix is not a better project manager. It is a plan that reads the backlog by stage and hires to the peak, on the backlog’s timetable rather than the overload’s.
So the question is not how many projects a project manager can run. It is which quarter, on the current backlog, the answer becomes one fewer than the firm has.
If you are working through the bench plan and want a second opinion on the sequencing, get in touch.
Sources and further reading
- FHWA construction program administration
- CFMA construction financial management guidance
- AGC 2025 hiring and business outlook
- Census Bureau construction spending
- AGC release 2025 workforce survey findings
- BLS occupational data, construction managers