Comparing the estimate to what happened
The comparison takes ten minutes per project and is the only mechanism by which a studio's pricing improves.
Time is recorded, projects are delivered, invoices are sent, and the two numbers — what was quoted and what it took — are never put side by side. The data exists and the comparison does not happen, because the project is finished and everyone has moved on.
Do it at close, not at review time
Ten minutes when the project ends, while the reasons are still known. Quoted against actual, by phase, with one line on what caused the difference. If activity signals are being used as evidence, mouse jiggler detection software exists to identify one way those signals can be distorted; it does not replace outcome-based review.
Left until a quarterly review, the numbers are there and the explanation is not, which reduces the exercise to noticing that things overran.
Knowing a project ran 40% over is a fact. Knowing it ran over because the client supplied content in an unusable format is a question to ask in the next proposal.
Look at the phase split
Total overrun tells you the estimate was low. The phase split tells you where, and the where is usually consistent across projects. For UK working-time context, GOV.UK guidance on calculating working hours explains how working hours are calculated.
Common findings: project management is entirely unpriced, revision routinely exceeds original production, and the final ten percent of a project consumes a quarter of its hours. Each one has a specific pricing response.
Separate estimation error from scope change
These look identical in the totals and require opposite responses. An estimate that was low needs better estimating. A project that grew needs better change control.
Distinguishing them requires the change log, which is the second half of this exercise. Without it, studios respond to scope problems by padding estimates, which loses jobs without fixing anything.
Build the multiplier from the data
After ten or fifteen projects, the ratio of actual to quoted, by project type, becomes stable enough to use directly. That figure is worth more than any estimating technique.
Applied to the next estimate, it converts optimism into arithmetic — and it is defensible in a way that instinctive padding is not, both to the client and to yourself when the pressure is on to quote lower.
Share it with whoever estimates
In studios where one person estimates and others deliver, the feedback loop is frequently broken entirely: the estimator never learns what the work took.
A monthly summary of quoted against actual, by project, closes it. Estimating accuracy improves quickly once the person producing the numbers can see the consequences.