Skip to content

Q&A · Agencies and outsourcing

Our fixed-price project went over budget again. How do we find out where the money went?

We hear this one from agency owners and delivery directors almost word for word. Here is how we’d start the investigation.

A tower crane rising against an empty sky over a construction site.
Fixed price, open sky: the quote is set, the costs are not. Photo: Wikimedia Commons, CC0.

Nythrex team · 10/29/2026

The question

Asked by an agency owner · the way we usually hear it

“We quoted a fixed price, delivered, the client is happy — and we lost money. Again. The timesheets say the team worked roughly what we estimated. The margin says otherwise. Where do I even start looking?”

Short answer

Put every cost of the project in one place, week by week: people’s time, AI, CI and cloud. Then compare it with the scope as it actually shipped — tasks and changes — not as it was quoted. Overruns are rarely one big mistake. They are many small additions nobody priced.

You are not alone, and it gets worse with time

The best-known numbers come from McKinsey and the University of Oxford, who studied more than 5,400 IT projects. Large projects — budgets above $15 million — ran on average 45% over budget and 7% over time, and delivered 56% less value than predicted. Each additional year of schedule increased cost overruns by 15%, and 17% of projects went so badly that the overrun exceeded 200%.

Your projects are smaller, so the percentages won’t transfer directly. The mechanism does: the longer a project runs, the more unpriced work accumulates, and the less anyone remembers why.

Cranes and workers on a dense city construction site between high-rise buildings.
Many small moving parts, one budget. Photo: Wikimedia Commons, CC0.

Where fixed-price margin usually goes

Common margin leaks and how they show up
LeakHow it shows upWhere to look
Small requests absorbed without a change orderTasks created after kickoff that were never in the quoteTracker, by creation date
ReworkSeveral pull requests and CI runs for the same taskRepository and CI, by task
Tooling that grew with the projectAI and CI bills that rose but were never split by clientProvider bills, by repository
Environments left runningCloud cost that continues after deliveryCloud cost, by project
Rounded timesheetsHours that don’t match any work in the periodTime next to tasks and changes

What we’d do before the next quote

  1. 1

    Freeze the quoted scope

    Write the quote down as a list of tasks. Everything added later is visibly different.
  2. 2

    Label additions as change requests

    Even the ones you decide not to bill. You can’t negotiate scope you never counted.
  3. 3

    Look weekly, not at the end

    A weekly cost per project shows drift while there is still time to talk to the client.
  4. 4

    Find the expensive changes

    Cost per shipped change shows which parts of the work — integrations, a fragile module, one environment — ate the margin.
  5. 5

    Quote from actuals

    The next estimate starts from what similar work really cost, including AI, CI and cloud, not from memory.

Where Nythrex helps — and where it doesn’t

  • What it does

    Brings the cost of a project into one view: tasks from Jira, Trello, Linear or Azure DevOps, pull requests, CI, AI from Claude Code, Codex and GitHub Copilot, and cloud from Azure or Google Cloud, with an optional estimate of people’s time. Views by project, team and period; every figure shows its basis.
  • What it doesn’t

    It doesn’t replace your invoicing or contract tools, and it doesn’t write your quotes. Cloud is counted at project level, not split across individual changes.

See how a project’s cost is assembled

Frequently asked questions

Why do fixed-price software projects go over budget?

Usually through many small, unpriced additions rather than one big mistake: client requests absorbed without change orders, rework, tooling costs that grew but were never split by client, environments left running and rounded timesheets. Longer projects accumulate more of them.

How often do IT projects run over budget?

In McKinsey and Oxford’s study of more than 5,400 IT projects, large projects ran on average 45% over budget and 7% over time, and each extra year of schedule increased overruns by 15%.

How do I find where a project’s margin went?

Put people’s time, AI, CI and cloud costs for the project in one place by week, compare them with the scope as shipped, and look at the cost of individual changes to find the expensive parts.

Stop guessing.
Start seeing.

No card and no subscription: paid plans have not launched yet.

Create accountNo card, no subscription