The Silent Margin Killer: Why Blended Delivery Is Replacing Pure In-House Teams
An all-employee Delivery team has long looked like the safe choice for margin. You control quality and keep the knowledge, in the process avoiding paying a partner's markup.
Yet the data suggests otherwise; the margin risk in an in-house model rarely shows up on a project report. It shows up in the hours between projects.
THE NUMBERS DON'T ADD UP
The 2026 SPI Professional Services Maturity Benchmark contains a contradiction worth stopping on. Project margins reached a five-year high of 37.7%. In the same year, billable utilization fell to 66.4%, the lowest level in the benchmark's 19-year history and well below the 75% SPI has historically considered healthy. Also average EBITDA sat at 9.9%, short of the 15% SPI expects.
Projects are making money. The businesses running them are making much less than they should.
A project can hit its margin target while the team that delivered it loses money.
The gap between those numbers is capacity: people who are employed, paid and available, but not billing. Project reporting doesn't see it, because it isn't attached to a project.
WHERE THE MARGIN ACTUALLY GOES
In a pure in-house model, four costs tend to sit outside project reporting:
Capacity built for peaks, paid through troughs. Demand in services arrives unevenly. A team sized to cover the busiest quarter carries that cost through the quietest one.
Specialists hired for occasional work. A skill needed on three projects a year still costs twelve months of salary. Between those projects, the specialist is either idle or doing work a less expensive person could do.
Ramp time. A hire made to meet a demand spike often arrives after it. SPI data published by Unit4 put the average time for a new consultant to become productive at nearly 64 days, with SaaS and software organizations taking more than 84. Every departure restarts that clock.
Staffing whoever is free. When people are on the bench, the pressure is to put them on the next project, whether or not they're the best fit. The cost shows up later as an overrun, and it usually gets recorded as scope creep.
What the reporting shows | What's often happening |
Non-billable time | Paid capacity with no demand behind it |
Scope creep | A skills mismatch on the staffing plan |
Growth investment (new hire) | Two to three months of cost before contribution |
Specialist billed on a project | Premium salary absorbed by lower-value work |
THE REAL COST OF AN INTERNAL HOUR
The usual objection to partners is price: a partner hour costs more than an internal one. That comparison is only true at full utilization.
An internal consultant's true cost per billable hour is their loaded cost divided by their utilization. At the industry's current 66.4%, every billable hour carries roughly one and a half times its loaded cost.
Billable utilization | Effective cost per billable hour (loaded cost = 100) |
80% | 125 |
75% | 133 |
66.4% | 151 |
60% | 167 |
A partner's rate is higher per hour, but the business pays nothing for a partner hour that isn't needed. The fair comparison is the partner rate, plus the internal time spent managing the partner, against the internal effective cost, not the internal salary.
THE PULLBACK THAT POINTS THE WRONG WAY
SPI also found that the subcontractor share of revenue eased in 2025, from 10.9% to 10.4%.
One reading of that is reassuring: firms needed partners less. Another is less comfortable. In the year utilization hit a record low, the industry as a whole pulled back on the variable part of its capacity while the fixed part stayed largely in place. When demand softens, partners are the easiest cost to remove. They are rarely the cost that should go first.
This is the shift underway. It isn't that partners are taking over Delivery. It's that the argument for an all-employee model, which assumes internal hours are always the cheaper option, no longer holds once idle time is counted.
BLENDED DOESN'T MEAN SMALLER
None of this argues for handing work away. The core team still matters most for the work that shapes renewals: solution design, customer relationships and adoption.
The change is in how that core is sized. The alternative is to size the core for baseline demand and treat partners as the planned way to absorb peaks and rare specialisms. Deciding which workstreams sit where is its own discipline, covered in depth in SDA partner Precursive's guide to the partner and internal mix.
THREE QUESTIONS FOR THE NEXT HEADCOUNT REQUEST
The clearest place to apply this is the hiring decision, before the fixed cost is committed:
Is this hire for baseline demand or for a peak? If the work disappears in two quarters, the role will become bench time.
How many weeks a year is this skill needed? If the answer is well short of a full year, a partner relationship may cover it at lower total cost.
What is the effective cost of the hire at realistic utilization? Compare that figure, not the salary, against the partner alternative and the time needed to manage it.
CONCLUSION
Low utilization is easy to treat as a scheduling problem. Often it's a structural one: a Delivery model that pays for capacity whether or not there is demand for it.
Blended delivery won't fix poor planning on its own. But it turns part of a fixed cost into a variable one, and in a market where a third of paid time goes unbilled, that difference shows up directly in margin.
Comments