Why do fixed-price projects lose money even when nothing breaks?

Because the price was set from a best-case plan and the hours past it arrive in small, ordinary pieces: longer status calls, a new stakeholder’s questions, internal reviews nobody estimated, a project that runs weeks past its date. On a fixed price the firm pays for every one of those hours. Splitting a closed project’s overrun into estimate miss, scope creep, calendar stretch and approved changes never billed shows which part a client conversation could have recovered.

What is the difference between scope creep and a change request?

The work can be identical; the route is different. A change request reaches a price, a date and a written yes before the hours are spent. Scope creep is the same work done without that decision, so it lands under the original project code at the original fee. A scope change is an approved change request: the SOW and the fee moved together.

How do you calculate the cost of scope creep?

From the time tracker. On each closed fixed-fee project, find the hours on outputs, rounds or stakeholders the SOW did not list, with no change request, and multiply them by each role’s loaded hourly cost from payroll. Keep them apart from hours past budget on work the SOW did list; those are an estimate miss, fixed in the next proposal rather than with the client.

How do I track project profitability when my team works across clients?

Code every hour to a client and a project, or to an internal code; set one loaded cost rate per role; charge outside costs to the project code. Each month, fee minus labor cost minus outside costs gives contribution by project, and grouping the rows by client shows which clients earn. Each week, compare hours used with hours budgeted for the current phase of every open fixed-fee project.

How do you measure agency profitability?

By project and by client, from the firm’s own records: contribution and effective rate, meaning fee divided by hours, on every closed project, grouped by client, with overhead left out of the first pass. An outside benchmark cannot say which of the firm’s clients earn; the time tracker and the invoices can.

Why Fixed-Fee Projects Lose Money: Count the Extras

How small untracked extras add up on a fixed-fee project

Nobody on the team decides to give work away. The extras are ordinary parts of client service, each too small to raise on its own, and each one lands somewhere in the time records — or nowhere:

The weekly status call that runs long, with the whole account team on it, every week the project runs past its planned end date.
A new contact on the client side who needs the background again: a re-brief meeting, a refreshed deck, a second walkthrough of work already approved.
Feedback that arrives late and in pieces, so a designer or writer reopens the same file several times instead of once.
Internal reviews that were never in the estimate: the creative director’s pass, the account lead’s pass, the partner’s read before anything goes out.
Small asks that sound like favors: one more size for social, a version for the board, a quick summary for a meeting the client is running.
Hand-over time when someone leaves the project halfway and the next person reads in.

Each of these is logged to the project code, to an internal code, or not at all. Illustrative example: a 25-person agency sells a website for a fixed $60,000, estimated at 400 hours, so each estimated hour is worth $150. The project closes at 520 hours. No time entry over the life of the project is longer than three hours, and the post-mortem finds nothing that went wrong. The fee is the same, and each hour actually worked earned about $115. The figures are round and hypothetical; the firm’s own time tracker gives the real ones.

Scope creep vs. a change request: where the line is

Scope creep and a change request can describe identical work. The difference is the route. A change request is added work that reaches a decision — a price, a date, a yes from someone with authority on each side — before the hours are spent. Scope creep is the same work done without that decision. A scope change, as owners use the term, is a change request that was approved: the SOW and the fee moved together.

Federal fixed-price contracts write the route down, which makes them a useful model even for a firm that never sells to the government. The FAR’s standard changes clause, 52.243-1, lets the contracting officer order changes “within the general scope” of the contract in writing. If a change moves the cost or the time of the work, the contracting officer “shall make an equitable adjustment in the contract price, the delivery schedule, or both,” and the contractor “must assert its right to an adjustment under this clause within 30 days” of receiving the written order. The version for architect-engineer and other professional services adds a sentence an agency could pin above every desk: “No services for which an additional cost or fee will be charged by the Contractor shall be furnished without the prior written authorization of the Contracting Officer.”

For an agency, three questions sort any piece of extra work, in this order:

Is it an output, quantity or round the SOW lists? Then it is in scope however long it takes. Hours past budget on it are an estimate miss, not scope creep, and the fix belongs in the next proposal, not in a conversation with this client.
Does it add an output, a round, a channel, a stakeholder or weeks to the calendar? Then it is a change, whether or not anyone calls it one.
Was there a written yes, with a price and a date, before the hours went in? If so, it is an approved change request. If not, it is scope creep, and the hours sit under the original project code at the original fee.

The wording of the reply to the client, and who on the account sends it, is in the guide on how to tell a client something is out of scope. The clause quoted above governs federal contracts only; it is quoted as published on acquisition.gov on October 22, 2026 (FAC 2026-01), while the FAR is being overhauled. What binds a private client is the firm’s own master agreement and SOW, and their wording is a question for the firm’s attorney. Not legal advice.

How to count what scope creep cost from time records

Take every fixed-fee project closed in the last twelve months. For each one, put four records side by side: the SOW and the estimate behind it (fee, hours by phase and role, outputs and rounds listed); the time entries (date, project, phase or task, role, hours, note); the change requests issued, with approval dates and amounts; and the invoices. Then sort the overrun into four kinds, because each has a different owner and a different fix:

Estimate miss: hours past budget on outputs and rounds the SOW listed. Owner: whoever estimates. Fix: the next proposal.
Scope creep: hours on outputs, rounds, channels or stakeholders the SOW did not list, with no change request. Find them in task names and entry notes, in work dated after a phase was approved, and in revision rounds past the count the SOW sold. Owner: the account lead. Fix: the change route.
Calendar stretch: project management, status meetings and internal reviews in the weeks after the planned end date. Owner: the project manager. Fix: a timeline clause.
Priced but not billed: change requests the client approved that never reached an invoice. Owner: finance. Fix: the invoice line opens on the yes.

Then put a cost on each kind. The cost of an hour is the role’s wage plus what the firm pays on top of it — payroll taxes, health coverage, retirement, paid time off — taken from the firm’s own payroll register. For scale, the Bureau of Labor Statistics put private employers’ cost in the professional and business services industry at $41.55 an hour worked in wages and salaries and $18.24 in benefits in June 2026, $59.80 in all. Illustrative example, continuing the website above: of the 120 hours over budget, 50 were an estimate miss on the build, 45 were scope creep — two extra page templates and a third review round for a new client stakeholder, none of it priced — and 25 were calendar stretch, project management and status calls in the three weeks the project waited on the client’s content. One approved change request, for $3,200, never reached an invoice. At a loaded cost of $70 an hour, the 120 hours cost the firm $8,400; at the $150 the hours were sold at, they were $18,000 of work. The figures are hypothetical. The split is the useful part: only the 45 hours of creep and the unbilled $3,200 were money a conversation with this client could have recovered, while the estimate miss and the stretch are fixed in the next SOW.

How to see profit by project and by client when the team works across clients

In an agency the same strategist can touch several clients before lunch, so profit by project cannot come from payroll, which pays people, not projects. It comes from the time tracker, and it needs three habits first: every hour coded to a client and a project, or to an internal code; one loaded cost rate per role, not per named person; and the outside costs of each project — freelancers, contractors, stock, media, project-specific software — charged to its code when the bill arrives.

Two rules keep the numbers honest. Work that serves several clients at once, such as a shared template or a training session, goes to an internal code instead of being spread by guess. And overhead — rent, admin, the owner’s time on the business — stays out of the first pass, because spreading it across projects by formula can bury the difference the table is meant to show. Once a month, the table is one row per project, grouped by client:

Fee: for a closed project, the fee as invoiced, approved changes included; for an open one, the fee beside the share of budget hours used, by phase. How revenue on an unfinished project is recognized in the books is a question for the firm’s CPA.
Hours by role from the time tracker, and labor cost: hours times each role’s loaded cost.
Outside costs charged to the project.
Contribution: fee minus labor cost minus outside costs. Effective rate: fee divided by hours.
Hours logged to an internal or general code that the entry notes tie to this client: work that never reached the project.

Grouped by client, the rows answer a different question than one project does. A client with three projects inside budget and one deep overrun is a conversation about one SOW; a client whose every project runs over is a conversation about the relationship, or about the price. The monthly table reports by role and client, never by named person. And every Monday, for each open fixed-fee project, the project manager posts three lines: hours used against hours budgeted for the current phase, whether that phase’s milestone is approved, and any change request still waiting for an answer. The owner reads only the projects where a phase has used its hours before its milestone was approved. That is the week to send a change request, or to decide on purpose to absorb the hours and say so in writing.

What to change in the next proposal

The count from closed projects is the raw material for the next SOW. Six changes follow from it, each a line in the proposal or a habit before it goes out:

Estimate from closed projects, not from the plan. Before pricing, pull actual hours by phase from the three to five most similar closed projects and start from those. In the Buehler study, the optimistic bias disappeared for participants told to connect their past experiences with the new prediction, and observers predicting someone else’s finish time erred the other way, overestimating it. So the hours get a second read from someone who will not run the project.
Carry project management, status meetings and internal reviews as their own lines: hours per week times planned weeks, by role.
Put the calendar in the price. State the planned end date, how many business days the client has to return feedback, and what happens to the fee when approvals push the project past the date — for example, project management billed at a stated weekly amount from the first week past it. The wording is the firm’s attorney’s.
List outputs with quantities, rounds per stage, the named approvers, and what is not included.
Write the change route into the SOW: a change request with a price and a date, approved in writing before work starts, and a separate time code that opens on the yes.
Phase the price when the scope is not yet definite: a discovery phase priced on its own, then a fixed fee for the build once the outputs are known. The FAR draws the same line from the other side: a time-and-materials contract “may be used only when it is not possible at the time of placing the contract to estimate accurately the extent or duration of the work.”

Then close every fixed-fee project with a fifteen-minute review: the four kinds of overrun in hours, written into the estimate library the next proposal starts from. A fixed fee is not the problem. A fixed fee priced from a plan instead of from the firm’s own history is.

When not to hire anyone, and where the Review fits

If closed projects land close to their budget hours, or the overruns are estimate misses that shrink from one quarter to the next, nothing needs fixing beyond keeping the count. If the count shows scope creep and calendar stretch, the fix costs little beyond attention: a change route, a timeline clause, an estimate library and the Monday check. Contract wording belongs with the firm’s attorney and revenue recognition with its CPA.

Fixed-fee margin sits at the last of the six hand-offs from a first inquiry to collected cash — payment to margin, what the firm keeps — and the same records can show leaks at the others: retainers past their hours, invoices past terms, proposals parked at “Sent.” If the pattern looks wider than one spreadsheet, the Revenue Review takes two weeks. It reads time entries, project budgets, change requests and invoices on read-only access under an NDA, opens no client work, reports time by role and client rather than by named person, and ends in a written report with a monthly dollar figure wherever the records support one and a 90-day plan; any build afterward is scoped and priced separately.

It is not worth buying when the count is small or the fix is already obvious, and the free 20-minute call is where we say which case applies.

Why Do Fixed-Fee Projects Lose Money Even When Nothing Goes Wrong?

Why a fixed fee loses money when nothing goes wrong