Hourly or Fixed Fee? The Five Documents That Decide Which One Costs You Less

A specialty food importer with about forty employees decided to replace the order management system it had outgrown, and hired an integration consultancy that quoted an hourly rate along with an estimate of roughly three weeks of effort. The engagement letter ran two pages. It named the rate, the invoicing cycle, and the estimate, and it described the work as configuration and data migration for the new platform. Nine weeks later the invoices had passed the estimate by a wide margin, the historical order data was partly loaded, and nobody on either side could point to a document that said what finished looked like.

Nothing dishonest happened. The consultancy billed hours it had genuinely worked, and the importer's operations manager had genuinely asked for each of the additional things that consumed them: a second pass at the product taxonomy, a customer credit-terms field that had not existed in the old system, reports rebuilt to match the ones the sales team already trusted. Each request was small. None of them entered the engagement through any process at all, because there was no process, and so the estimate quietly stopped describing the work while continuing to describe the reader's expectations.

Working backwards from that invoice, the decision that made it likely was not choosing hourly over fixed fee. Hourly was arguably the right model for a migration whose data quality nobody had inspected yet. The decision was accepting an hourly engagement with none of the instruments that make hourly legible: no scope document, no ceiling, no invoice format, no acceptance test, no change log. Fixed fee would have failed in its own way, by producing a fight over whether the credit-terms field was inside the price. The five checks below are what people who buy professional services repeatedly put in place before they care which model they are buying.

1. Is the scope a document, or is it the proposal email?

A proposal describes what a firm intends to do; a scope document describes what the reader is entitled to receive, which is a different sentence written by a different part of the brain. The working version lists deliverables by name, states what each one includes, and states plainly what it excludes. In the importer's case, a scope of two pages would have said that migration covered open orders and twenty-four months of history, that the product taxonomy was inherited as-is from the old system, and that custom reporting was not included. Every later dispute lived in one of those three lines.

The test on any engagement, hourly or fixed, is whether somebody who was not in the sales conversation can read the document and tell you what is owed. Give it to your controller, or your warehouse lead, and ask them to describe the finished state. If they cannot, the scope is not a scope. Buyers who do this often keep a standing exclusions paragraph they paste into every engagement, covering data cleanup, training, third-party license fees, and work caused by another vendor's delay, because those four items generate most of the overruns they have paid for.

2. Does the estimate have a mechanism attached to it?

An estimate with no mechanism is a mood. What turns it into a control is a not-to-exceed figure, a notification trigger set well below it, and a stated consequence when the trigger is hit. The usual construction: the firm notifies you in writing at seventy percent of the estimate, work above the not-to-exceed number requires your written approval before it is performed, and unapproved hours above the ceiling are not billable. That last clause is the one that matters, because a notification duty with no penalty produces notifications that arrive with the invoice.

The importer had none of this, and the consultancy had no obligation to raise a flag, so the first signal that the estimate had broken was an invoice arriving after the money was spent. Experienced buyers also ask for the estimate broken into phases with its own ceiling per phase, which converts one large unmanaged number into three or four small ones that fail visibly and early. Fixed-fee engagements need the mirror instrument: a written statement of what happens to the price if the assumptions underneath it turn out to be wrong, and which specific assumptions those are.

3. What is the invoice required to show before you pay it?

Most hourly disputes are really invoice-format disputes. A bill that reads "professional services rendered, 61.5 hours" cannot be checked, cannot be allocated to a project code, and cannot be argued with. The format worth specifying in the engagement letter has one line per task per person per day, with a narrative entry describing the task in language a non-specialist recognizes, the timekeeper's name and role, the rate applied to that role, and the increment used. Six-minute increments and quarter-hour increments produce measurably different totals on work that arrives in short bursts, and the difference is worth naming in writing.

Ask also for the rate card by seniority, and for a rule about staff substitution, because the quiet cost escalator on hourly work is the mix shifting toward senior people for tasks a junior could do, or toward a junior on work you hired the principal for. The Federal Trade Commission oversees how businesses represent prices and terms to their customers, and the practical reflection of that at your desk is simply this: whatever the firm told you about who would do the work belongs in the paperwork, with a rate beside each name. Firms that bill this way are usually glad to; their timekeeping systems already produce it.

4. Who decides the work is finished, and against what test?

Outcome pricing sounds like it removes the argument about hours and it does, replacing it with an argument about completion, which is worse if undefined and better if defined. The instrument is an acceptance test written before the work starts: a short list of conditions that, when demonstrated, oblige you to sign off and oblige the firm to stop billing. For the migration, that list would have been concrete. Open orders reconcile to the old system's totals. Two years of history is queryable. Three named users can place an order end to end without assistance. Sign-off within five business days of demonstration, or a written list of what failed.

The importer's engagement had no acceptance test, which is why week nine felt indistinguishable from week four. Nobody could say whether they were near the end, because the end was not a place. Buyers who have run many of these keep acceptance criteria in the same document as the deliverables and tie payment milestones to them, so that the money moves when something demonstrable happens rather than when a calendar month closes. That single link, milestone to demonstration, does more to keep an engagement honest than the choice between hourly and fixed ever does.

5. How does new work get in, and at what price?

Every one of the importer's overruns entered the project through a hallway conversation, which is the normal way work enters a project and the reason a change log exists. The mechanism is unglamorous: a numbered running list, one row per request, with the date, who asked, what it changes in the scope, the estimated hours or fixed price, the effect on the schedule, and the approval. One person on each side holds the pen. Requests that are not on the log are not in the engagement, and everybody knows that before it matters rather than after.

The same log is what makes hourly and fixed fee comparable in practice, because after two or three engagements it tells you what proportion of your projects arrive as changes. A buyer whose logs are consistently short can take fixed fees with confidence, since the scope holds. A buyer whose logs run long is buying discovery work, and should pay by the hour with a ceiling rather than pay a firm to price uncertainty it cannot measure either. The paperwork does not just control the current job; it tells you which model to ask for next time.

The importer eventually finished the migration, on a second engagement structured in three phases with a ceiling each, an acceptance test per phase, and a change log that the operations manager updated on Friday afternoons. The rate did not change and the firm did not change. What changed was that the work became something both sides could see the edge of, which is the whole function of those five documents, and the reason people who buy professional services often stop treating the hourly-versus-fixed question as the important one.