Blog
7 min read
A RIBA stage fee is a fixed price for a bounded block of design work, agreed before the work starts. Learning how to track time against RIBA stage fees means booking every hour to the stage that consumed it, then setting that total beside the fee you quoted. Most practices manage the first half of that sentence. Very few manage the second.
So the same pattern repeats. Stage 4 runs hot. Someone notices in month three, by which point Stage 5 has begun and is quietly soaking up the overflow. The fee for the next project comes off the same spreadsheet as last time, because nothing in between produced a better number.
Because you fix the price early and discover the effort late.
The RIBA Plan of Work 2020 divides a project into eight stages, running from Stage 0 strategic definition through to Stage 7 use. Practices price each stage as a share of the whole appointment. Stage 4 technical design usually carries the largest share, and its scope is also the hardest to draw a line around.
Then there is the accounting gap. Your percentage split hands money to stages. Your timesheet, offering only a project field, hands hours to nothing smaller than the whole job. So the two numbers never meet. You can know a project sits at 60% of its fee with 40% of the drawings issued, and still have no idea which stage did the damage.
Margin leaves little room for that. Deltek's 46th annual Clarity study covered nearly 700 architecture and engineering firms across the US and Canada, and put operating profit on net revenue at a ten-year high of 21.4%. Treat that as the ceiling. Then take a stage that overshoots by a third of its fee: it eats well into the margin on the job.
The odd part is that billing already respects the stage boundary. Neumann Monson's account of how architects structure fees describes practices breaking a fixed fee down by design phase for billing, and charging hourly in the early phases while the scope is still loose. The invoice knows what a stage is. The timesheet does not.
The next fee proposal stops being a guess.
Suppose Stage 3 took 340 hours on a school you finished last year, and the fee assumed 260. Eighty hours out. That is an argument. You can lift the Stage 3 percentage on the next school, or narrow what Stage 3 includes, or do both and say why. A practice without that number repeats 260 and hopes.
Nothing about this is peculiar to architecture. It is the same discipline that sits behind designing profitable projects in creative agencies. Margin targets hold there for one reason: the scoping numbers come from finished work rather than from optimism. Different profession. Identical failure.
There is a second benefit, and it lands sooner. Stage-level time makes an overrun visible while the stage is still running. That is the only window in which anybody can act on it.
Give every stage its own budget line, book hours to that line as they happen, and review the burn weekly. Weekly, not monthly.
Five steps cover it:
Nothing there needs new software. It does need a timesheet with a stage field. That rules out more of them than you would expect.
Leaks in a stage budget are predictable. Which means you can watch for them by name.
| RIBA stage | What the fee buys | Where hours go missing |
|---|---|---|
| Stage 1 preparation and briefing | Project brief, feasibility, initial programme | Unbilled advice before appointment |
| Stage 2 concept design | Concept proposals, outline strategies | Option studies nobody agreed to |
| Stage 3 spatial coordination | Coordinated design, planning submission | Planning officer correspondence |
| Stage 4 technical design | Technical information for construction | Consultant coordination and RFI churn |
| Stage 5 manufacturing and construction | Site queries, inspections, design responses | Site visits and contractor queries |
Stage 5 is worth singling out. Modest fee, long duration, and queries that arrive one at a time over months. No single response feels billable. The total is.
Stage 1 has the opposite problem. Much of the work happens before an appointment exists, so there is no project code to book it to. The hours never appear anywhere.
Open the code at the first meeting instead. If the job never converts, you have at least learned what pursuing that kind of client costs the practice in unpaid hours. Worth knowing on its own.
Four, and they fit on one line per project.
When hours left falls below forecast hours to complete, the stage is going over. That is the whole test. It shows up well before anyone would have felt it.
Record the overrun, name the cause, then decide who carries it.
The cause matters more than the number. It is one of three things: the brief moved, the stage was priced too thin, or the team worked slower than the estimate assumed.
Each one leads somewhere different. A moved brief is a change request. A thin price is a proposal correction. A slow team is a resourcing conversation. Only the timesheet tells you which of the three you are having.
An overrun that traces back to a vague deliverable is a proposal problem wearing delivery clothes. The fix sits upstream, in how you define deliverables so a client cannot expand them. Give "coordination drawings" a count and an acceptance test and you have something you can size a stage budget around. Something a client cannot widen once Stage 4 is under way.
Do not roll the overrun silently into the next stage. That is how a bad Stage 3 becomes a bad Stage 4, and how a practice ends a project unable to say which stage lost the money.
One obstacle here is cultural, and it is bigger than the software one. A stage that overruns looks like somebody's mistake, so the instinct is to absorb it and say nothing. Say plainly that the variance is data rather than a verdict. Otherwise a team that expects to be blamed for the number will find ways to make the number look fine. Then the whole exercise measures nothing.
Two projects of the same type give you a range. Five give you a percentage split worth defending.
The mechanics are dull and that is the point. Keep a table of completed stages with the hours each one took, the fee it carried and the building type. When a new appointment comes in, read the split off the table rather than off the last proposal. Adjust for what you know is different. Write down what you adjusted for, so the next reader can follow the reasoning.
Expect the table to contradict you. The interesting finding is rarely that a project lost money overall. It is that one stage subsidised another inside a job whose final account looked healthy. A total landing near the fee hides that. Only the stage-level split shows which half of the appointment was earning.
Korrel works this way round: proposals carry deliverables and milestone billing, and time and cost then track to those estimates as the job runs. The point is not the tool. It is that the stage budget and the stage hours live in the same place, so the comparison happens without anyone building it.
Start with the stage you are running right now. Split its fee into hours, ask the team to book to it for the rest of the stage, and see what the number says at completion. One stage will not settle the argument about whether timesheets earn their keep. It will tell you whether your Stage 4 percentage has been wrong for years.
COMMON QUESTIONS:
RELATED READING:
Set timesheet timing from the decision the data must support. Use this freshness policy to keep staffing, margin, billing, and estimates current.
Design the boundary of each stage before work starts. Define its output, included review, dependencies, and trigger for reopening the fee.
Separate the refused submission from the new response. Map each refusal reason to tasks, hours, costs and supporting evidence before you issue a priced change.
Korrel turns briefs into structured proposals, then tracks what each job actually costs.
START FREE TRIAL