Cost overruns don't start mid-project: they start at the quote

A cost overrun almost never starts where you notice it. By the time a budget turns red mid-delivery, the overspend was usually baked in weeks earlier, at the quote and in the plan. One of the reasons I joined Teamwork.com is because I'd spent years in agencies watching profitable-looking projects quietly lose money.

This guide covers what a cost overrun is, how to calculate one, and where overruns actually begin. It names the model I rely on, the Quote-to-Cash Overrun Trail, and shows how to protect margin as AI changes what delivery costs.

What is a cost overrun?

Cost overruns often get treated as a mid-project shock, when the number was usually settled long before delivery started. That gap rarely appears out of nowhere. It was priced in.

A cost overrun is the amount by which a project's actual cost exceeds its approved budget. For example, if you budget $9,000 and spend $12,000, your cost overrun is $3,000.

You'll also see it called a budget overrun, a cost increase, or budget overspend. They all describe the same gap between your estimate and your actual cost. The label matters less than where the gap was born.

A cost overrun is not the same as cost escalation. Cost escalation is a rise you can see coming, for example inflation pushing up contractor rates over a long engagement. A cost overrun is the unplanned gap you never priced for in the first place.

The distinction that matters most in services is cost overrun versus change order. A change order is extra scope the client has formally agreed to pay for, so it protects your margin. A cost overrun is extra work you delivered but never repriced, so it comes straight out of your margin instead.

The services version of this problem has its own signature. In construction or manufacturing, an overrun usually shows up as materials, equipment, or subcontractor costs you can point at. In services, the overspend is almost entirely labour, so it hides inside timesheets and capacity rather than on a purchase order. That's why an overrun can run for weeks before anyone outside the delivery team feels it.

I care about that difference because it decides who pays. When I frame absorbed scope as a cost overrun rather than goodwill, teams start repricing it instead of quietly eating it. The same hours, named honestly, change the conversation.

How do you calculate a cost overrun?

In some delivery team I've been part of, the hours turned red long before any finance report did. That's why I treat hours as the earliest proxy for a cost overrun: hours are where your money actually goes. The formulas are simple, and I keep both close, because one tells you the size of the gap and the other tells you its severity.

$$\text{Cost overrun} = \text{actual cost} - \text{budgeted cost}$$

$$\text{Cost overrun \%} = \frac{\text{actual cost} - \text{budgeted cost}}{\text{budgeted cost}} \times 100$$

Say a typical fixed-fee project is scoped at around 120 hours and priced at $15,000, with a blended cost rate of $75 an hour. Your budgeted cost is $9,000, so you've planned a $6,000 margin, or 40%. This is an illustrative scenario, but the shape holds across most fixed-fee delivery.

Now delivery runs long and the team logs 160 hours. Actual cost climbs to $12,000, a $3,000 overrun, or 33% over budget. The client still pays the agreed $15,000, so those 40 extra hours add no revenue; they come off your margin.

That $3,000 is the whole story. Your margin falls from $6,000 to $3,000, so a 33% cost overrun just halved your profit on the job.

The percentage is what tells you how worried to be. A $3,000 gap on a $9,000 budget is serious; the same $3,000 on a $90,000 budget is noise. I read hours against budget as variance analysis worth checking weekly, and I treat every overrun as a hit to margin control, not just a line that went over.

Hours give you that signal earliest, because you can read them daily. Cost and margin follow once the timesheets settle, but by then you're reacting. So I convert the budget into an hours ceiling on day one, and treat any drift past it as the overrun starting, not a rounding error.

Why cost overruns start at the quote (the Quote-to-Cash Overrun Trail)

I've watched the same shape play out on fixed-fee projects again and again: margin loss usually starts before delivery, when teams quote from optimism and plan from vague briefs. The overrun only becomes visible at invoicing, but it was decided far upstream. That failure has a shape, and I call it the Quote-to-Cash Overrun Trail.

Stage 1: Quote

The estimate carries all the risk. Price from optimism, or from a rival's number instead of what comparable work actually cost, and you've committed to a margin you may never see. A quote is a promise about hours you haven't spent yet, so a loose one is an overrun waiting to happen.

Stage 2: Plan

Scope and resourcing bake the overrun in. A loose statement of work, no contingency, or the wrong people assigned turns an ambitious quote into an impossible one. The plan is your last cheap chance to catch a bad number before hours start burning against it.

Stage 3: Deliver

Hours burn against the budget. This is where most guides tell you to watch a live budget, and you should, but by now the overrun is being confirmed, not created. Watching delivery alone is like reading a smoke alarm and calling it fire prevention.

Stage 4: Invoice

You recover, reprice, or absorb. If the quote and plan were sound, this is where margin shows up. If they weren't, this is where it quietly disappears, one unbilled hour at a time.

Read the trail end to end and the pattern is clear. Overruns are born at stages one and two, caught at stage three, and paid for at stage four. Fix the stage where the number goes wrong, not the stage where it finally hurts.

There's a 2026 twist that sharpens all of this. AI is compressing delivery time, and on hourly or time-and-materials billing, every hour AI saves comes off the invoice, not the cost base. Finish faster and you bill less for the same outcome, so a healthy-looking quote erodes even when the work goes well. The billing model you chose, whether fixed-fee, time-and-materials, or retainer, decides who carries that risk.

This is why the billing model has quietly become a margin decision, not just a commercial preference. On a fixed fee, AI efficiency can widen your margin, because you keep the difference between the price and the lower cost. On time-and-materials, the same efficiency hands the saving to the client unless you reprice around outcomes. I'd rather choose that trade-off deliberately than let the invoice choose it for me.

Why do project budgets really blow up?

Most overrun advice says watch your budget more closely, and I think that treats a symptom. The budget going red is the overrun surfacing, not the overrun happening. The real causes sit upstream on the trail, and when I map them back to where they start, the same five keep appearing.

Optimism bias is the biggest single cause, and it isn't a services problem alone. Bent Flyvbjerg's research on large projects found that 91.5% of projects go over budget, over schedule, or both, with a mean cost overrun of 62%. Estimating from gut feel instead of grounded cost estimation is the fastest way a quote goes wrong at stage one.

The pattern holds wherever you look. According to PMI's 2018 Pulse of the Profession, 43% of projects across industries are not completed within budget. In Wellingtone's 2026 State of Project Management, only 36% of organisations mostly or always complete projects on time.

Scope creep is the cause that bleeds margin most quietly, because it rarely arrives as one big ask. It shows up as a run of small unbilled favours: an extra revision here, a quick call there, a deliverable nobody scoped. Each feels too minor to raise a change order for, and together they rewrite the economics of the job.

The projects that lost me money were rarely the disasters. They were the ones that felt fine until the final margin came in. That's what makes scope creep so dangerous: it never trips an alarm.

Resourcing and utilisation gaps do similar damage earlier on the trail. Put too few people on the work, or the wrong seniority mix, and the hours inflate before anyone notices a problem. A team quietly running above a healthy utilisation target is an overrun forming in plain sight.

The AI-compressed cause is the newest, and it's the one teams underprice most. Bill by the hour and a faster delivery simply means fewer billable hours for the same scope. The cost base barely moves, but the revenue drops, so the margin you quoted shrinks even when quality is high. I treat that as a pricing problem to solve at the quote, not a discount to hand over at the invoice.

The tracking gap is what keeps all of this invisible. Promethean Research's 2026 agency profitability analysis found about 41% of digital agencies don't track individual project margins, so overruns stay invisible until the P&L shows them. That matches our own Sprint to AI research: only 1% of teams can manage data, projects, profits, and resources in a single tool. If you can't see margin forming, you can only react once the hours are already spent.

How do you spot a cost overrun before it lands?

By the time a budget report turns red, I've usually already spent the hours it's warning me about. Lagging signals tell you the money's gone; leading signals tell you it's about to go. The trick I rely on is to act on the leading ones, which is why I keep them in what I call the Overrun Early-Warning Matrix.

The leading signals all live in your delivery data, as long as that data is connected to your budget. Hours are the earliest proxy, so I reforecast to completion the moment actuals start to drift. The whole point is to buy back time you'd otherwise lose.

The 40%-complete mark is the one I watch hardest. It's early enough that you can still reforecast and reallocate, yet late enough that the numbers are real rather than noise. If actual hours are outpacing the plan at that point, the finish will almost always overrun unless something changes. Waiting for the halfway budget report just costs you the room to react.

Pro tip: Catch drift while you can still act on it. Real-time budget tracking shows planned versus actual hours and cost as delivery happens, so a project trending over surfaces early.

A matrix only helps if you turn it on your own live work. So before you assume a project is on track, I'd run it through a quick check rather than wait for the month-end number to make the decision for you.

How to stop cost overruns before they reach the invoice

You can't prevent every overrun, but I've learned you can stop most of them being priced in. Prevention is really estimating and planning discipline applied before the quote goes out, not heroics during delivery. These four steps are the ones I lean on hardest, and I run them in order.

Step 1: Estimate from history, not optimism

Pull hours, cost, and margin from comparable past projects before you price anything. Then quote and cost the work deliberately so the number you commit to reflects what similar delivery actually took. This is the highest-leverage move, because it fixes the overrun at stage one.

Step 2: Run formal change control

Decide up front what counts as new scope, and make a change order the default response rather than an awkward conversation. Every quick favour you absorb is an overrun you chose. A clear line between goodwill and billable scope protects the margin you quoted.

Step 3: Track against a live budget

Watch planned versus actual hours as delivery happens, so drift is visible while you can still act on it. The alternative is finding out at month-end, when the hours are already gone. A live view turns a lagging signal into a leading one.

Step 4: Plan resources and utilisation on purpose

Model capacity with resource management before you commit, not after the work is already slipping. Overruns often start as the wrong people on the wrong work, or a team quietly running hot. Getting the seniority mix and the workload right is prevention, not admin.

What to do when an overrun happens anyway

When an overrun lands anyway, the instinct I keep running into is to eat it quietly and move on. That's the most expensive option, because you lose the money and the lesson at the same time. A clean recovery has four steps, and I run them in order too.

Step 1: Find the root cause

Trace the overspend back along the trail: was it the quote, the plan, the scope, or delivery? Fixing the wrong stage just repeats the overrun next quarter. The honest answer is usually further upstream than it feels.

Step 2: Reprice or reallocate

Decide whether the extra scope becomes a change order, a rebalanced team, or a cost you knowingly absorb. Make the call while hours still remain, not after the invoice. A decision you make on purpose always beats one made for you by the calendar.

Step 3: Communicate early

Tell the client before the invoice, not with it. A mid-project conversation protects the relationship, while a surprise line item damages it. Clients forgive a timely heads-up far more readily than a shock bill, so I never let the invoice deliver the news.

Step 4: Capture the learning

Feed the real hours and final margin back into your estimating model, so the next quote for similar work is sharper. An overrun you learn from stops being a loss and becomes a calibration. The data is only useful if you actually keep it.

What does a cost overrun look like in practice?

The overruns that cost me most were rarely disasters; they were the projects that felt fine until the final margin came in. I've seen this play out most clearly on fixed-bid work, where the whole risk sits in one number. Two illustrative scenarios show how a healthy-looking bid turns thin.

Picture a fixed-bid municipal IT contract, scoped and priced at a round $90,000. You'd costed it at about $60,000, so you planned roughly $30,000 of margin, near 33%. A few weeks in, the real state of the infrastructure appears, and meeting the spec needs far more work than the RFP implied.

The extra work adds around $20,000 in cost, pushing actual cost to $80,000. Your margin falls to about $10,000, near 11%, so an underscoped quote just erased two-thirds of the profit. Nothing went wrong in delivery; the overrun was priced in at the bid.

The same trap catches agencies on creative work. Picture a rebrand quoted at a fixed fee that then absorbs three unbilled rounds of revisions. The hours that made it profitable are gone, without a single change order raised.

The reason I keep two very different scenarios side by side is that they share a root, not a symptom. One is a hard infrastructure surprise; the other is soft, social scope creep on creative work. Yet both trace to the same place on the trail. Both would have been caught by the same discipline: estimating from real comparable hours and pricing contingency into the bid.

Both scenarios are hypothetical, but the mechanism is real and repeatable. In each case the delivery team did nothing careless. The margin was lost at the quote and the plan, exactly where the trail says it starts.

How Teamwork.com keeps overruns off your margin

I spent years stitching this together from spreadsheets: the quote in one file, the hours in another, the margin arriving weeks after the work was done. I'd hand Teamwork.com to any delivery lead still living that way. It connects the quote, the plan, the hours, and the invoice in one system, which is the difference between catching an overrun while you can still act on it and reading about it in the P&L.

• Start at the quote. Cost management software lets you price from comparable actuals instead of a hopeful guess, then track cost and margin against that number as the work runs. That's stage one fixed before a single hour is logged.

• Then test the promise before you make it. Model the work with Tentative Projects before a deal closes, so you know you have the capacity to deliver at the margin you quoted. That's the plan stage caught early, when it's still cheap to fix.

• Then watch the hours burn. Planned versus actual hours update as delivery runs, so stage three confirms the margin you quoted instead of quietly spending it. That's the difference between reading a smoke alarm and preventing the fire.

• Then let AI price the risk. TeamworkAI projects margin before you commit with the AI Profitability Forecaster, and utilization insights show where capacity runs hot or idle. The AI Teammate Kash handles costed financial work as a supervised line item with an owner, so faster delivery stays a repricing decision you control.

The proof holds up. When SugarCRM unified projects, time tracking, and billing in Teamwork.com, they reached near-perfect invoicing accuracy, crediting less than $20K on more than $10M in annual invoicing. That's stage four closing cleanly, where margin so often leaks.

Put together, that's the Quote-to-Cash Overrun Trail closed end to end. The quote, the plan, the hours, and the invoice finally share one view, so an overrun has nowhere left to hide.

The four truths behind most cost overruns

After enough of these, I stopped treating overruns as bad luck and started seeing four repeatable truths. I keep them in mind every time a new quote goes out.

• Overruns are inherited, not incurred: The quote and plan hand delivery a margin it can only confirm or lose.

• Visibility beats vigilance: You protect margin by connecting the numbers, not by watching a single budget harder.

• AI changes the maths: On hourly billing, efficiency and overrun read the same, so your pricing has to adapt.

• Evidence sharpens the next quote: Every real overrun is data your estimating model should absorb.

The cost overrun questions clients and PMs ask most