Fixed Bid vs T&M vs Capped T&M: How We Actually Choose

Every agency founder eventually picks a side in the pricing war. Fixed-bid people think T&M shops are lazy. T&M people think fixed-bid shops are gambling. After a decade of shipping client work, we think both camps are half right — and the actual answer is a three-way choice, made per engagement, with rules you can write down.

This is how we decide.

The three models, stated honestly

Before the framework, the definitions — including the parts vendors usually skip.

Fixed bid. You quote a number for a defined scope. The client pays that number whether the work takes you 200 hours or 600. Your margin is your problem. Their scope discipline is also your problem, because every "small tweak" eats your float.

Time and materials (T&M). You bill hours at a rate. The client pays for what they get. Your margin is predictable per hour. Their budget is unpredictable, which means their finance team is uncomfortable, which means you are uncomfortable by month three.

Capped T&M. You bill hours at a rate, but agree a ceiling. Under the cap, it's T&M. Over the cap, it's your problem — or you renegotiate. It looks like the best of both worlds on a slide. It isn't, but it's often the right answer.

What none of them solve

No pricing model fixes a bad discovery, a client who can't make decisions, or a team that doesn't track time. If you're losing money, the contract structure is probably the third or fourth thing wrong. Fix the others first.

The decision framework we actually use

We ask four questions, in order, before quoting anything.

1. How well-defined is the scope? Not "do we have a brief" — do we have wireframes, acceptance criteria, and a clear definition of done?
2. How stable is the client's business context? Are they pivoting, fundraising, or about to reorg?
3. How much novel technical risk is there? New integration, unfamiliar stack, AI components with unclear evaluation?
4. What does the client's procurement actually allow? Some enterprises literally cannot sign T&M without a cap.

Map the answers like this:

Scope	Context	Tech risk	Recommended
Tight	Stable	Low	Fixed bid
Tight	Stable	Medium	Capped T&M
Loose	Stable	Any	T&M with sprint reviews
Any	Unstable	Any	T&M, short engagement
Tight	Stable	High	Capped T&M with risk buffer

The table isn't gospel. It's a starting point that stops the conversation drifting toward whatever the client asked for first.

Where each model actually shines

Fixed bid: marketing sites, well-defined integrations, repeat patterns

Fixed bid works when you've built this thing before. A Shopify theme customization, a marketing site on a known stack, a Stripe integration into an existing app. You have a reference implementation, the scope fits on one page, and the client understands that "fixed" means "fixed scope, not fixed wishes."

It also works when the client genuinely needs budget certainty and is willing to trade flexibility for it. Government work, grant-funded projects, anyone whose finance team thinks in line items.

The trap: estimating fixed bid on novel work. We've watched smart founders quote a fixed number on "build us an AI agent for X" because the client wanted certainty. The client got certainty. The agency got a 40% margin haircut.

T&M: discovery, R&D, long-running platform work

T&M shines when nobody can honestly say what "done" looks like. Discovery phases, prototype-then-decide engagements, platform teams embedded for two quarters. The client gets transparency. You get fair compensation for actual effort. Nobody pretends a Gantt chart is real.

The trap: T&M without a cadence. If you bill monthly with no demo, no burn-down, and no "here's what we'd cut if budget tightens" conversation, the client will eventually wake up, audit the invoices, and decide you've been milking them. Run T&M like a product team runs sprints: visible work, visible decisions, visible burn.

Capped T&M: the negotiated middle

Capped T&M is what we end up signing roughly half the time. The client needs a not-to-exceed number for their board. We need protection against scope creep. Both sides accept that the cap is a ceiling, not a target.

The contract language matters here more than people realize. A capped T&M clause should specify:

- The cap amount and what it covers (scope statement attached)
- What happens at 70% and 90% burn (mandatory written notice)
- The change-control process for raising the cap
- What happens if work completes under the cap (client keeps the difference; we do not bill the cap)
- Out-of-scope work is paid T&M, no cap, signed change order required

Without the 70/90 notice clause, you will absolutely deliver bad news on the last day of the project. Don't do that to yourself.

The numbers behind the model choice

We model every engagement two ways before quoting: as fixed bid and as T&M. The gap tells us where the risk sits.

A quick example. Say internal estimate is 480 hours at a $150 blended rate. T&M revenue: $72,000. We'd typically quote fixed bid at a 25–40% buffer over the internal estimate, depending on confidence — so $90k to $100k. Capped T&M lands in the middle: cap at $84k, bill actuals.

If the client pushes hard on fixed bid for novel work, the buffer goes up, not down. In our experience, novel scope overruns the central estimate by 30–60% surprisingly often. If your buffer is 15% on a project with three unknowns, you're not pricing — you're hoping.

A small spreadsheet beats a big argument

We keep a one-tab model per project: estimated hours per workstream, confidence (low/med/high), buffer percentage by confidence, blended rate, and a comparison row. Five minutes to fill in, ends every internal debate about whether a quote is "too high." If finance and delivery disagree, they disagree on a cell, not on a feeling.

War story: the fixed bid we should have refused

A few years back we took a fixed-bid contract to build a "simple" data pipeline between a client's CRM and a warehouse, plus a dashboard. Scope looked tight on paper. We quoted with a 20% buffer.

What the brief didn't say: the CRM had been customized by three different consultancies over six years, half the field names were Hungarian-notation leftovers from a 2018 migration, and the client's "single source of truth" was actually four sources that disagreed politely.

We shipped. We lost money. The client was happy, which was the only saving grace.

The lesson wasn't "don't do fixed bid." The lesson was that we hadn't done discovery before quoting. We took the client's framing of "simple" at face value. Now, for anything touching legacy data or third-party systems we don't own, we quote a paid discovery first and refuse to bid the build until discovery completes. That's a model choice — discovery is small T&M, build is whatever the discovery tells us it should be.

If you want the discovery-first pattern in more detail, we've written about it under our delivery playbook.

What we tell clients who push back

Clients ask for fixed bid because they want risk transfer. That's reasonable. The honest answer is: we can transfer the risk, but you'll pay for it in the buffer, and you'll lose flexibility to change your mind. Most clients, when you put it that way, choose capped T&M.

Clients ask for T&M when they're sophisticated, internal, or burned by a previous fixed-bid disaster. The honest answer is: yes, and here's the cadence and reporting we use so you never feel like you're writing a blank cheque.

Clients who insist on fixed bid for genuinely undefined work are telling you something about how they'll behave during delivery. Believe them. Either bid with a buffer that reflects the actual risk, or walk.

Where we'd start

If you're an agency founder still defaulting to one model, do this next week: pull your last ten projects, mark each as profitable or not, and write the actual pricing model next to each. The pattern will be obvious within an hour. Then write down — literally, on one page — the rule for when you'll use each model going forward. Share it with sales. Stop negotiating the structure on every call. Negotiate the number.