Fixed fee vs capped T&M: choosing the commercial model per workstream

Agencies often debate fixed fee versus time and materials as if it were a company-wide philosophy. Clients prefer fixed fees because they want certainty. Agencies prefer T&M because they want protection. The debate goes in circles because it treats an engagement as one thing, when most engagements contain several kinds of work with very different levels of uncertainty.

A better question is: for each workstream, who is in the best position to carry the risk of the unknown? That answer usually changes within a single project.

Fixed fee rewards agencies that estimate well and control scope. It punishes agencies that let requests in informally, because every extra hour comes straight out of margin. Capped T&M shares the risk until the cap, then behaves exactly like fixed fee. Pure T&M protects the agency but makes clients nervous, and nervous clients scrutinize every invoice.

Capped T&M is often sold internally as the safe middle ground. In practice, caps have a gravitational pull. Clients treat the cap as the budget, and work expands toward it. Once burn approaches the cap, the remaining work is effectively fixed fee, with the added problem that the client has watched the hours tick up all along.

Consider a $150,000 capped engagement where the team has burned $120,000 by the 60% milestone. Two things can happen. Either someone notices at that point and raises the forecast with the client, or nobody notices until the cap is hit and the final 40% of the work is delivered unpaid. The model did not protect the agency. The weekly burn check would have.

• Report burn against the cap every week, alongside percentage of deliverables complete.
• Agree in the SOW that the agency will notify the client at 75% of cap, with a forecast.
• Define what happens at the cap: work pauses, a change order extends it, or remaining scope is reprioritized.
• Track out-of-scope requests separately even under T&M. They still consume the cap, and the client should know which hours went where.

Most of the healthiest SOWs we see during onboarding use more than one model. A typical software engagement might look like this:

Clients understand this structure quickly when each section explains why. "We have fixed the price where we can see the work clearly, and capped it where we depend on systems we have not seen yet" is a sentence most procurement teams appreciate. It also tells them exactly where the risk sits.

Every model needs change control. The trigger is what differs.

Fixed fee

The trigger is a request outside the listed deliverables, exclusions or assumptions. The evidence is the SOW clause. The mechanism is a change order with a price. This is where informal chat requests do the most damage, because nothing in the commercial model signals that a line was crossed.

Capped T&M

The trigger is a forecast that the cap will be exceeded, or a request that changes the direction the cap was based on. The evidence is burn data plus the original assumptions. Out-of-scope requests matter here too: if the cap was set for one integration and the client adds a second, the cap should move.

Pure T&M

There is no price trigger, but there is still a priority trigger. New requests displace planned work, and the client should see the trade-off. A monthly summary of what was requested, what was done and what was deferred keeps T&M relationships healthy.

Procurement teams often push for a single fixed price because it is easier to approve. A mixed model can still be presented as one number: the fixed-fee workstreams plus the caps, shown as a total maximum. The client gets a ceiling for budget approval, and the agency keeps the risk allocation it needs. Add a short table showing which workstreams are fixed and which are capped, with one line on why for each.

If procurement insists on fixing everything, fix it with a contingency line rather than by inflating every estimate. A visible contingency, released only through change control, keeps the conversation honest and gives the PM a clear mechanism when the uncertain workstream turns out as uncertain as expected.

1. Can you list the deliverables and count them? If yes, fixed fee is viable.
2. Does effort depend on systems, data or people you do not control? If yes, cap it rather than fix it.
3. Is the goal to learn rather than to produce something specific? If yes, use T&M or a fixed-fee timebox.
4. Do you have weekly visibility of burn against plan? If not, neither fixed fee nor caps will protect you.

The last point is the one that decides outcomes. In TWINJAM's onboarding cohort, the biggest margin losses were not tied to the commercial model chosen. They were tied to whether anyone was comparing the work being done to the contract while the project was still running. Margin telemetry and scope monitoring are what make any model work as designed.