Most architecture projects that go over budget go over for three reasons, and they show up in nearly every post-mortem we have sat through. The schematic design hours were estimated at proposal time, when the firm knew the least. Revisions after an approved milestone were absorbed instead of billed. And construction administration was budgeted at the traditional 20% of the fee, then consumed 25% to 30% of the effort. There is a fourth reason, and it is the quiet one: nobody saw the overrun until the invoice, because the hours lived in a spreadsheet that got updated monthly.
This page is the hub for everything we write about efficiency and project management. It defines the problem, walks through the data that exists (and the data that does not), and points you to the posts that go deeper on each cause and each fix.
What does “over budget” mean on an architecture project?
Two different budgets get blown on a building, and they get confused constantly. The client’s construction budget is the cost of the building. The architect’s fee budget is the hours the firm can spend and still make money. This page is about the second one, though the first causes plenty of damage to the second: when bids come in 18% high, somebody has to redesign, and that somebody is usually working for free.
The fee budget is arithmetic. Take the net fee (what is left after consultants are paid), divide by a blended billing rate, and you have hours. A $180,000 net fee at a $150 blended rate is 1,200 hours. Split those by phase and you have a budget. Spend 1,380 hours and you are 15% over.
What that 15% does to profit is the part principals underestimate. Deltek Clarity 2026 treats a net labor multiplier of 3.0 or higher as the mark of a high-performing A&E (architecture and engineering) firm, and puts median overhead at about 161% of direct labor. Run the multiplier math from the AIA (American Institute of Architects) Handbook of Professional Practice: 160% overhead gives a break-even multiplier of 2.6. Spend 15% more hours than planned on a project priced at 3.0 and its effective multiplier drops to about 2.6. The whole profit is gone, and the next overrun comes out of the firm.
How often do architecture projects run over budget?
Honest answer: there is no clean published survey that says what share of architecture projects exceed their fee budget. Anyone quoting one is quoting a vendor slide. What the profession does publish is the result at the firm level, and it is consistent year to year.
Deltek Clarity 2026 reports operating profit on net revenue at 16.7%. The AIA Firm Survey Report 2024 counts roughly 19,000 to 20,000 architecture firms in the United States, about three in four with fewer than 10 people, and finds that the stipulated (fixed) fee is the most common way those firms get paid.
Put those together. Most firms work on fixed fees, so an overrun lands on the firm, not the client. Average profit is mid-teens. The distance between average and high performer is roughly five points of margin, which on $3 million of net revenue is $150,000 a year, or about 1,000 hours of write-offs at a $150 rate. That is where the overruns live: not in a survey, but in the gap between firms that catch drift in week three and firms that find it on the invoice.
Why are schematic design hours always optimistic?
SD (schematic design, the phase where the building’s form, plan, and systems get decided) gets the traditional 15% of the basic fee. On the 1,200-hour project above, that is 180 hours: about four and a half weeks for one person. It covers a site visit, a code summary, two massing options, plans and elevations, a cost check, and one client presentation.
It never covers what SD becomes. The client wants a third option. The third option is a hybrid of the first two. The planner asks for a shadow study at the pre-application meeting that was not in the proposal. A 24-unit multifamily in a coastal zone picks up a design review hearing and two more renderings. Somewhere in there the team falls in love with a scheme, which is good for the building and bad for the budget.
The root problem is that the fee is set at the moment the firm knows the least about the project, and SD is where the unknowns get paid for. Three fixes work. Price options explicitly: two in basic services, more as an additional service. Budget SD by task before you sign, not by percentage after. And stop the SD clock with a written approval; the phase-based budgeting post explains why that signature matters more than any other piece of paper on the job.
Why does construction administration blow the fee?
CA (construction administration, the phase after the permit when the building actually gets built) gets 20% of the fee in the traditional split the AIA Handbook describes: SD 15%, DD (design development) 20%, CD (construction documents) 40%, bidding 5%, CA 20%. Ask any firm that tracks phase hours and they will tell you CA regularly consumes 25% to 30% of the effort. That is practitioner consensus rather than a survey figure, and we have not met a principal who disputes it.
The reasons are structural. CA is priced by percentage but consumed by duration. A TI (tenant improvement) with a six-week permit clock and a four-month build is a different animal from the same fee on a project where the contractor takes 14 months. Every extra month brings more site visits, more RFIs (requests for information, the contractor’s written questions), more submittals, and more pay application reviews. None of them are optional.
AIA B101-2017, the standard owner-architect agreement, saw this coming. Section 4.2.3 lets you fill in the number of site visits, submittal reviews, and RFI responses included in the fee, beyond which the work is an additional service. Most firms leave those blanks empty, or fill them in and never invoice against them. The scope creep post covers how to do it without a fight; the short version is that the letter is easy to write in month two and impossible in month eleven.
Where do unpriced revisions come from?
Every set has revisions. The expensive ones come after a milestone approval. The owner signs off on DD, the team starts CDs, and three weeks later the owner’s new head of operations wants the loading dock on the other side of the building. Under B101-2017 Section 4.2.1, a change in the owner’s instructions or approvals is an additional service. In practice the PM (project manager) says “sure, we can look at that,” and 60 hours disappear.
The second source is the bid. Construction pricing comes in high, the owner cannot fund the gap, and the firm redesigns to a lower number. Whether that is billable depends on the cost-of-the-work language in your contract (B101 Article 6) and on whether anyone updated the construction budget when the program grew. Usually nobody did.
The third source is internal. A principal redlines a set the night before it goes out, and the redlines are not corrections; they are a different design. The rework post puts numbers on what a redline costs. The point here is that internal revisions are invisible to the client and to the budget, because nobody codes them as anything but CDs.
The fix is dull. Keep a written record of what was approved at each milestone, keep a change log, and hold a firm rule that any change after approval gets an hours estimate before it gets a yes. Not every change gets billed. Every change gets counted.
Why don’t firms see the overrun until it is too late?
Because the data arrives late and lives in the wrong place. Timesheets get filled out Friday afternoon from memory. The budget lives in a spreadsheet the PM updates when the invoice goes out, so the first real comparison of hours to fee happens a month after the hours were spent. Then the invoice waits: Deltek Clarity 2026 puts average DSO (days sales outstanding, the time from invoice to cash) at about 68 days.
Add it up. Work done in early March gets entered in mid-March, compared to budget in April, invoiced in April, and paid in late June. By the time the number is undeniable, the phase that produced it is over and the team has moved on to the next one.
Utilization, oddly, is not the problem. Deltek Clarity 2026 puts firm-wide utilization (the share of paid hours charged to projects) just under 60%. The hours are being worked. They are being worked on the wrong things, or on the right things after the budget for those things ran out two weeks earlier.
What actually keeps an architecture project on budget?
The firms that stay on budget are not better estimators. They see drift earlier and act while the action is still cheap. In practice that is six habits.
- Load the phase budget before kickoff. Hours by phase and by role, in whatever tool the team will actually open. A budget that lives only in the proposal is not a budget.
- Compare hours to percent complete weekly. If CDs are 40% drawn and 55% of the CD hours are gone, you know in week four instead of week twelve.
- Run a 15-minute status meeting with the numbers on the screen. Budget, schedule, blockers, next seven days. No design review.
- Write the additional services letter the day the change happens. Use the 4.2.1 and 4.2.3 language you already signed.
- Price CA by duration and count, not by percentage. Months of construction times site visits per month, plus an RFI and submittal allowance.
- Close every project with a post-mortem that changes a template. Otherwise the same overrun repeats on the next job with a different address.
None of this needs a big firm. Most of it needs a PM with 2 hours a week and a number she trusts.
Where to go next
Every post in this pillar goes deeper on one cause or one fix. Start with the one that matches the pain you have this week.
- Why Do Architecture Projects Always Run Over Budget? (And What the Data Says): this page. Bookmark it; everything below hangs off it.
- What Is a Good Utilization Rate for an Architecture Firm?: benchmarks by role, and why a principal at 80% is a warning sign.
- How to Stop Scope Creep Before It Eats Your CA Fee: B101 Article 4 in plain English, and when to send the letter.
- Can Architects Actually Work a Four-Day Week?: what the pilot data says and what it takes in a deadline business.
AECPulse Project Manager tracks each phase budget against logged hours and flags a project the week it starts drifting, which is the difference between a course correction and a write-off.