How to reduce cost overruns on projects already in flight

Most advice on cost overruns is really advice about the next project. Estimate better, add contingency, tighten the SOW. All true, and none of it helps the project that's 40% delivered and 55% spent right now.

This guide is about that project. You'll get a weekly check I call the Burn Gap Triage, three moves for closing the gap, and the delivery habits that stop small leaks becoming write-offs.

What actually counts as a cost overrun on a services project?

The textbook definition misses the part that hurts services teams most. A cost overrun is the amount by which a project's actual cost exceeds its approved budget. For example, a $60,000 fixed-fee project that costs $72,000 to deliver has a $12,000 overrun, and on fixed fee, every dollar of it comes straight out of your margin.

On services projects the cost is almost entirely people's hours, so the overrun builds one timesheet at a time. Who absorbs it depends on the billing model. You eat it on fixed fee, the client usually does on time and materials, and on a retainer it hides as overservicing. If you want the full breakdown of causes and the formula, our guide to cost overrun causes and the overrun formula covers the pre-project side in depth.

Why overspend hides until the month-end close

Picture a retainer where everything looks fine on the dashboard until the last week of the month, when the timesheets finally land. That's the pattern on most services teams I've been part of: overspend happens daily, but it only becomes visible monthly. The gap between those two moments is the detection lag, and it's where most overruns are really born.

Before I joined Teamwork.com, I spent nine years agency-side, running retainers on a basic timer plus spreadsheets. By the time a spreadsheet told me an account was over, the hours were spent and the client already had the extra work. The only conversation left was an awkward one.

The lag comes from a handful of predictable places:

This is my Detection Lag Map. Read the middle column again. Almost nothing on it becomes visible in the same week it happens.

In Teamwork.com's 6 Strategic Shifts for 2026 research, 27% of leaders said clients moving budget mid-project is their biggest frustration. That frustration is really a detection problem. A client shifting priorities isn't the overrun; the overrun is the three weeks your team keeps delivering against the old plan before anyone updates the budget.

The favor problem works the same way: scope creep rarely starts with an unreasonable request. It starts with something small and well meant that never gets written down. Our guide on how to handle scope creep and protect your margins covers the change control side in detail. On retainers, the same pattern shows up as overservicing on client accounts.

If any of this sounds uncomfortably familiar, it's not a sign you're bad at your job. It's a sign your numbers arrive late. Fix the timing, and most of the judgment calls get easier.

The Burn Gap Triage: a weekly check that catches overspend early

If the problem is lag, the fix is a shorter feedback loop. The Burn Gap Triage is a 15-minute weekly check that compares how much budget you've spent with how much of the job you've actually delivered. The difference between those two percentages is your burn gap, and it tells you whether a project is heading for an overrun long before the budget runs out.

This isn't a new idea dressed up. It's a plain-language version of earned value management, the method project controls teams have used for decades. A PMI paper on practical earned value calculation notes that research has found EVM warning signals reliable as early as 15% into a project. The difference is that you run it weekly, on every active project, with numbers your team already produces.

Step 1: Measure budget burned

Start with money, not hours, if your people have different cost rates. Budget burned is the cost of all time and expenses logged to date, divided by the total project budget. On a $60,000 project with $33,000 of logged cost, you've burned 55%.

The number is only as good as your timesheets. If half the team logs time on Friday afternoon from memory, you're measuring last week's guesses. That's why weekly timesheet approval sits at the heart of this check, not on the side of it.

Pull the figure from the same place every week, using the same project cost tracking setup. A budget vs actual tracker template works if you're starting from scratch, but the goal is a number nobody has to rebuild by hand.

Step 2: Measure scope delivered (honestly)

Scope delivered is the share of agreed deliverables that are actually done and accepted, not "in progress". Weight deliverables by their planned effort, so finishing the logo doesn't count the same as finishing the website. The honest part matters. "Nearly done" is the most expensive phrase in delivery, because it lets a project report 80% complete for three weeks running.

Step 3: Read the gap against your threshold

The burn gap is a simple subtraction:

$$\text{Burn gap} = \text{Budget burned (\%)} - \text{Scope delivered (\%)}$$

A positive gap means you're spending faster than you're delivering. A small gap is normal noise, especially early on. The threshold is where you stop watching and start acting. Set it before the project starts, so nobody negotiates it down when the number turns red.

For example, take that $60,000 fixed-fee project in week 6 of 12. You've burned 55% of the budget ($33,000) but delivered 40% of the weighted scope. That's a 15-point gap. If the rest of the project runs at the same efficiency, your estimated cost at completion, in dollars, is:

$$\text{Estimated cost at completion} = \text{Budget} \times \frac{\text{Budget burned (\%)}}{\text{Scope delivered (\%)}} = 60{,}000 \times \frac{55}{40} = 82{,}500$$

That's a $22,500 overrun, spotted with six weeks left to do something about it. At month-end close, you'd have spotted it with two. To decide when a gap like that needs action, these are the thresholds I'd start with. They're my own working rule of thumb, so tune them to your margins:

Step 4: Trigger a move, not a meeting

A red reading has to change what someone does this week, otherwise the triage becomes another report people glance at and ignore. Agree in advance that every amber or red result produces a named owner, one of the three moves below, and a date.

Visibility is what makes this sustainable across a portfolio. When OIC Advisors brought their projects into one platform, they gained 360° visibility across all active projects with 100% less time spent manually generating reports. That's the difference between a weekly check that survives month three and one that quietly dies.

Recover, Recoup, or Reset: pick the move that fits the gap

Not every overrun is yours to absorb. The most common response to a red burn gap is "we'll just work harder". That's the most expensive fix available, because it burns more of the same hours you're short of. The triage tells you something's wrong; the cause tells you which of three moves to make.

That's the Three-Move Response Matrix. Most teams only ever use the first row.

Recover: fix the delivery leak

Recover is the right move when the client hasn't asked for anything extra and you're still over. Common culprits are rework from unclear briefs, senior people doing tasks a junior could handle, and too many internal review rounds.

Look at where the hours went, not just how many there were. A time report split by person and task usually shows the leak within ten minutes. If your most expensive person has logged 30% of the hours on work that was scoped for someone cheaper, that's your gap.

The fix is operational: reassign tasks, cut an internal review round, or tighten the brief for the next deliverable. Recover doesn't need the client, which is why it's tempting to use it for everything. Only use it when the cause is genuinely internal.

Recoup: bill for the scope you're actually delivering

Recoup is for the gap the client created, even politely. If you've delivered three rounds of revisions on a two-round contract, the fourth round is a change order, not goodwill.

The hard part is timing, not the paperwork. Clients accept change orders far more easily when they arrive within days of the request, before the work is done. The same pressure shows up in our research: 66% of leaders say clients are now more demanding but less willing to pay. That makes recouping early more important, not less.

Reset: re-baseline with the client before the gap becomes a write-off

Reset is the move for when the plan itself is wrong. The client moved the launch date, a dependency on their side slipped, or the project turned out to be a different shape from the one you quoted.

A Reset conversation doesn't start with "we're over budget". It starts with "here's what's changed, here are the options". Bring two or three choices: cut or defer scope, extend the timeline, or add budget. Letting the client choose keeps the conversation commercial rather than personal.

For example, take a $20,000-a-month retainer covering roughly 130 hours at a $150 blended rate. By week three, the team has logged 120 hours because the client added a campaign mid-month. A Reset offers the client three paths: move the campaign's second phase into next month, drop two lower-priority deliverables, or add a one-off $4,500 for 30 extra hours.

Most clients pick one of those options without fuss. What they don't forgive is finding out after the invoice that the retainer quietly ran 30 hours over and someone decided to bill it, or to swallow it, without asking. If you need a refresher on building the budget itself, our guide to creating and managing a project budget covers setup.

Seven habits that reduce cost overruns during delivery

Seven small habits do more to keep the burn gap small than any single heroic recovery. The three moves handle a gap once it appears. These habits make gaps smaller and rarer, week after week.

1. Log every favor the day it happens

A favor doesn't have to be billed, but it does have to be written down. Log it as time against the project, tagged as out of scope, even if you choose not to charge for it. Once favors are visible, you can decide on purpose which ones to give away. That turns overservicing into a relationship investment you can actually measure.

2. Put a 48-hour clock on change requests

Any request that changes scope gets a written response within two working days, with a cost and a yes, no, or "let's discuss". A change request that sits in someone's inbox for two weeks has usually been delivered by the time anyone prices it.

3. Close timesheets weekly, not monthly

Weekly approval turns timesheets from an accounting chore into a delivery signal. It also shrinks the detection lag from weeks to days, which is the whole point.

Make the approval part of the burn gap check, not a separate task. The delivery lead approves the week's time, then reads the gap. Same 15 minutes, one habit.

If people resist, show them the math. A timesheet filled from memory at month-end tends to lose small tasks and meetings, and those are exactly the hours that turn into invisible overspend.

4. Protect senior hours on fixed-fee work

On fixed fee, your margin depends on who does the work as much as how long it takes. An hour from your most senior person can cost two or three times an hour from a mid-level specialist. Set cost rates for every role in Teamwork.com's quoting and costing tools so that difference shows up in the budget, not just the timesheet.

Plan which tasks need senior judgment before kickoff, and treat any senior time beyond that plan as an amber flag. It's usually a sign the brief was unclear or the junior team needs support.

5. Standardize the status report so it stops eating unbilled hours

Client reporting is one of the biggest hidden costs on any account. On the accounts I ran, every client wanted something different: a live walkthrough for one, a deck for another. Every report started from a blank page, and none of those hours were billable.

A standard status format fixes two problems at once. It cuts the unbilled hours, and it gives you a regular place to show the client the budget position before it becomes a surprise.

Build the burn gap into the report itself. A client who sees "65% of budget used, 60% delivered" every two weeks is rarely shocked by a change order. Our guide to client reporting has more on structuring these.

6. Re-plan resourcing when the gap opens, not after

An amber gap is the moment to look at who's booked on the project for the next four weeks, not the moment the budget runs out. Moving one person off, or swapping in a lower-cost specialist for routine tasks, is often all the Recover move needs.

7. Keep a live change log for handovers

Account leads move on, and the budget history often leaves with them. A change log that records every scope decision, favor, and approved change means a new lead can pick up an account without rebuilding the story from memory. It also protects you in a Reset conversation. "Here's what changed and when" lands far better than "we think it was around March".

Where overrun recovery usually goes wrong

Even with the habits in place, recovery attempts fail in a few predictable ways. The surprising thing is that most failures come from good intentions, not bad management. These are the five I'd watch for.

Waiting for the month-end number to be sure. The instinct is to wait until the data is complete before raising a flag. By then the hours are spent. A rough weekly reading beats a perfect monthly one every time.

Asking the team to "just be more efficient". This sounds like a Recover move but isn't one. Without a named leak and a specific change, it's a request to work unpaid overtime, and it tends to show up later as burnout and rework.

Cutting quality to hit the number. Dropping QA or a review round saves hours this week and creates rework next month. If a quality step has to go, that's a Reset conversation with the client, not a quiet internal decision.

Re-estimating without telling the client. Updating your internal forecast is healthy. Keeping it from the client means the eventual conversation is about trust, not budget.

Treating the overrun as the project manager's personal failure. Overruns are usually a system problem: late data, no change control, no agreed threshold. Blaming the person running the project makes the next one hide its numbers for longer.

The overruns that hurt most are rarely the ones you got wrong at the quote. They're the ones everyone could see coming and nobody had the authority, or the data, to call early. The fix for most of these is the same. Agree the threshold, the owner, and the three moves before the project starts, so nobody has to invent a response under pressure.

What kind of tool actually helps you reduce cost overruns?

A tool helps you reduce cost overruns only if it shortens the time between spending money and seeing it. Most of the mistakes above trace back to data that arrives late or lives in three separate places. The setup you run delivery on decides how long that lag is.

I think of this as the Overrun Visibility Stack. The rule is simple: the further up the table you are, the more you have to assemble the burn gap by hand. The more manual it is, the less likely it happens every week.

Spreadsheets aren't the villain here. They fail quietly rather than suddenly, which is exactly why they're dangerous for budget tracking. A generic PM tool is great at tasks, but it can't manage money. A traditional PSA (professional services automation) platform handles the money, but only if people actually use it.

What you're looking for is budgets, time, delivery progress, and billing in one place that the delivery team is happy to work in every day. That combination is what makes a weekly burn gap check take 15 minutes instead of an afternoon.

How Teamwork.com keeps budget burn visible while you deliver

Imagine running the Burn Gap Triage without exporting a single spreadsheet. As an agentic PSA, Teamwork.com keeps budgets, time, and billing in one platform your delivery team actually wants to use, so the numbers are live when you need them. Here's how each part of the check maps to the platform.

The first thing the triage needs is a live budget number, and that starts with a budget type that matches the contract.

Budget tracking for fixed fee, time and materials, and retainers Know the moment spend crosses your line. Set up a budget per project, split it across tasks, and get notified at a threshold you choose. I'd set alerts at your amber point, so the warning arrives before the weekly check, not after it.

A budget alert tells you something's wrong, but a Recover or Reset decision needs the margin picture behind it.

Profitability reports and AI Forecaster See margin move while the project is still live: Profitability Reports show cost, revenue, and profit per project in real time. The AI Forecaster predicts where profitability will land based on your historical revenue and costs. That's the evidence you want in the room for a Reset conversation.

None of that works if the hours arrive a month late, which brings us back to timesheets.

Time tracking and timesheets Get hours logged in the week they happen: your team uses timers or bulk timesheets and marks time billable or not. Automated reminders nudge anyone who forgets. Timesheet approval gives the delivery lead a natural weekly moment to read the burn gap.

Once you've found the leak, a Recover move usually means changing who does what for the next few weeks.

Workload Planner Rebalance the team the moment an amber gap appears. The Workload Planner shows who's booked on what, so you can reassign tasks, swap in a lower-cost specialist, or free up a senior person.

The last piece is closing the loop to the invoice. When a Recoup move produces a change order, the extra time and budget flow through to invoicing, and invoices sync with accounting tools like Xero and QuickBooks. You're not re-keying hours into a separate billing system, which is where a surprising amount of recouped revenue quietly goes missing.

AI Teammates add another layer. Take the blank-page pain out of the standardized status report from habit five: Flo, one of Teamwork.com's AI Teammates, checks project health and generates client updates.

FAQ