Your utilisation looks fine. Your project margin is the one lying to you.
In IT and digital services, the cost that decides your margin is the cost nobody bills: the bench, the pre-sales hours, the scope that crept, the rework. A blended day rate hides all four. Time-driven costing puts a number on each, project by project and client by client, so the work you keep chasing is the work that actually pays.
Cost and Profitability Consulting · 150+ models since 2010 · TDABC
In IT and digital services, 25 to 40 percent of total cost is bench time and overhead that direct project hours never capture. Costing only billable hours overstates project margin and hides that 20 to 30 percent of clients are unprofitable once pre-sales, bench and rework are loaded. TDABC assigns those hidden hours to the work that caused them, so every project and client shows its real margin.
The hours that decide margin are the ones you never invoice.
A services P&L looks healthy at the top. Day rates are competitive, utilisation targets are met, the pipeline is full. Then the operating margin comes in thin and nobody can say exactly which engagement drained it. The reason is structural: a blended rate prices the billable hour and quietly absorbs everything around it, the bench between projects, the pre-sales a deal needed to close, the project-management overhead, the rework when scope moved. None of it appears on an invoice, so none of it appears against the project that caused it.
Spread those costs evenly and the well-run, fixed-scope engagement subsidises the messy one. The team is told the messy one is fine, because at a blended rate it looks fine. It is not the rate that is wrong. It is that the rate was never asked to carry the cost of complexity.
FROM CONTRACT FEE TO REALISED MARGIN
Illustrative. The contract fee steps down through billable delivery, bench allocation, unbilled pre-sales and scope creep. What a blended rate reports as margin is mostly the cost it forgot to bill.
Project margins run from minus 15 to plus 45 percent. Three things move them.
Scope creep
Present in roughly 38 percent of projects. Requirements drift, the build absorbs hours nobody re-quoted, and the overrun lands silently in delivery cost rather than in a change order.
Bench allocation
The single largest hidden cost, around 25 percent of the gap. Time between billable work is real cost; if it never lands on a project, every project looks cheaper than it is.
Rate discounting
Roughly 20 percent. The day rate on the contract is not the rate realised once discounts, non-billable ramp and write-offs are counted. Realisation, not the rate card, sets margin.
The lever underneath
Utilisation ties them together. A 5-point gain in billable utilisation typically lifts operating margin by 3 to 5 points, but only once you can see where the unbilled hours actually go.
Two projects, same day rate, opposite truth.
Two delivery engagements sign at the same contract value and the same blended day rate. On the monthly report they look like twins, and the second is often the one the team is encouraged to repeat. Trace the unbilled hours and they separate completely.
| Project A | Project B | |
|---|---|---|
| Contract value | €240,000 | €240,000 |
| Blended day rate | €1,050 | €1,050 |
| Reported margin | +22% | +24% |
| Scope changes | 2 | 11 |
| Rate realisation | 94% | 71% |
| True margin (TDABC) | +28% | −12% |
Project B is the one heading for a follow-on at the same terms. A cost model puts its real margin on the table before that decision, not in the year-end review when the cash has already gone.
ONE STANDARD PROJECT, BUILT FROM TIME EQUATIONS
Illustrative. Two parameters replace a thousand timesheets: a capacity cost rate per team, and a time equation that builds a project from scoping, design, development, testing and deploy. No surveys, and it updates as the work changes.
Cost lands on the work, not the average.
Set the capacity cost rate
For each resource group, cost per minute of practical capacity, usually 80 to 85 percent of the theoretical. Unused capacity then shows up as unused capacity, not as a heavier rate on everyone.
Write the time equations
Describe how a project consumes time: fixed scoping and deploy, plus hours per requirement, per story point, per support tier. The model builds project cost from drivers, not from memory.
Load the unbilled hours
Bench, pre-sales, account management and rework attach to the project and client that caused them, so the cost a blended rate hid finally has an owner.
Rank, then act
Projects and clients sort by true margin. The losers become candidates to reprice, re-scope, change how they are served, or set a minimum, each with the number behind the call.
CUMULATIVE PROFIT, CLIENTS RANKED BEST TO WORST
Illustrative. The whale curve in services is steep. A handful of well-run accounts build profit past 100 percent of the firm's total; the demanding, scope-changing tail gives a large share of it back.
Best in the room on data. Weakest on turning it into cost.
IT and digital services score higher than any other sector on Data & Technology: the timesheets, the tickets, the project tooling are already there. The gap sits on Cost Allocation and TDABC Process Design. The raw hours exist, but they are never turned into true project and client cost, so the firm flies on utilisation and gut feel where it could fly on margin. The fix is method, not more technology.
THE 7 COST DIMENSIONS · TYPICAL IT-SERVICES PROFILE
Illustrative profile, not a benchmark. The shape is consistent: strong on data and tooling, thin where the data should become per-project and per-client cost.
When code writes itself, what happens to the billable hour?
AI is reshaping the economics of this sector faster than any other. Code generation compresses delivery hours, AI agents absorb tier-one support, and the billable-hour model itself comes under pressure. When delivery hours fall, the firms that win are the ones that already price on value and cost-to-serve rather than on time. A cost model is how you make that move with numbers, not nerve: see which work AI makes cheaper to deliver, and reprice it before the market does it for you.
Price the outcome and its true cost, not the hour.
Firms that cost on real consumption stop repeating the engagements that drain them and stop discounting the accounts that quietly carry the firm. The same model feeds the proposal desk, so every bid goes out grounded in what the work will actually take, and the move from time-and-materials toward value pricing rests on evidence rather than hope.
The model is built on the data you already hold and handed to your team, so it stays current as your delivery mix and your tooling change.
Frequently asked questions
- How do you calculate the true cost of an IT project?
- Go beyond billable hours. Load bench allocation, pre-sales, project-management overhead and rework onto each project using time equations, costed per minute of practical capacity. A blended day rate alone understates project cost by 25 to 40 percent, because it never bills the hours that complexity actually consumes.
- Why are some IT-services clients unprofitable?
- Scope creep, low rate realisation and high account-management intensity. Once pre-sales, bench, account management and rework are fully costed, 20 to 30 percent of clients in a typical services firm lose money, even when the headline day rate looks healthy.
- What is TDABC for IT services?
- Time-Driven Activity-Based Costing assigns cost with two parameters: a capacity cost rate per resource group and time equations that describe how each project and client consumes time. It produces per-project and per-client margin without timesheet surveys, and it scales as your delivery model changes.
- How much does utilisation affect margin?
- A great deal, but only if you can see where the hours go. A 5 percentage-point gain in billable utilisation typically lifts operating margin by 3 to 5 points. The bench is not idle time to be tolerated; it is unpriced cost that has to land on the work that caused it.
Find out which projects quietly carry the firm.
The Profit Check takes five minutes and no data upload. It tells you where the distortion in your project and client cost is likely to be, and what it is worth to correct.