Your estimates are off by the same amount every time: measure your overrun factor

Projects do not run long at random. Take your last three finished projects, divide the hours you tracked by the hours you quoted, and the same number comes out each time: 1.25, 1.30, 1.35. That number is your overrun factor, and it is more honest than any estimate you will ever write. Multiply the next estimate by it. Then set the project’s budget to the hours you quoted, so the overrun shows up at hour 24 instead of on the invoice.

Why your projects always take longer than you quoted

Because an estimate is built from the plan, and the plan is the version of the project where nothing goes wrong. The client answers on the same day. The files are the right files. The third round of feedback does not exist. You are not lying when you estimate this way; you are describing the project as it would go if it went well.

Psychologists have a name for this and a number. In 1994, Buehler, Griffin and Ross asked students to predict when they would finish their honours thesis. The average prediction was 34 days. The average reality was 55, and fewer than a third of them finished by the date they had named. Kahneman and Tversky had described the pattern fifteen years earlier and called it the planning fallacy: people predict from the inside of the plan, where every step looks tractable, rather than from the outside, where similar projects have a track record.

Freelancers are not exempt. In freelancermap’s 2025 survey of 3,571 freelancers, 29 percent said they always or frequently go over the hours they planned for a project. Five percent said never.

The part that matters for a freelancer is not that the error exists. It is that the error is stable. You do the same kind of work, for the same kind of client, with the same habits, and the gap between the plan and the hours comes out about the same every time. A stable error is a measurable one, and a measured error is a pricing input.

Quoted 30 hours, tracked 39: what the overrun costs

The arithmetic is the division as always: fee over every hour the job took.

	Fee	Hours	Effective rate
As quoted	3,000	30	100
As tracked	3,000	39	77

Against a floor of 85, the project you quoted at 100 an hour paid 77. Nothing unusual happened. The client did not add scope. There was no disaster. It simply took 30 percent longer than the plan, the way the last two did.

This is a different problem from the fixed-price overrun where the fee runs out of hours. That guide gives you the limit, fee divided by floor, and tells you to check at the halfway hour. This guide is about the number before that: how many hours to write in the quote in the first place, given what your own projects actually do.

Three projects, one factor

Here are three finished projects, the hours quoted, the hours tracked, and the ratio between them.

Project	Quoted	Tracked	Factor
Landing page	20 h	25 h	1.25
Brand deck	30 h	39 h	1.30
Web app	40 h	54 h	1.35
All three	90 h	118 h	1.31

The plans vary. The factor does not. That is the whole finding, and it is the reason the fix is a multiplication and not a resolution to estimate better. Bent Flyvbjerg, who has spent a career on why large projects overrun, calls the method reference class forecasting: do not predict from the plan, predict from the outcomes of similar projects that already happened. For a freelancer with tracked hours, the reference class is sitting in the time log.

How to measure your overrun factor

Take the last three finished projects. Finished, because a project in progress has not shown its overrun yet. Three is enough to see the pattern; five is better if you have them.

Divide tracked by quoted, per project. The hours you actually logged over the hours that were in the quote. If you never wrote hours into the quote because you priced it as a fee, use the hours you had in mind when you set the fee. You had a number; that is the one.

Then divide the totals. All tracked hours over all quoted hours. In the example that is 118 over 90, which is 1.31. This is the factor to use, because it weights the large projects the way your income does.

Keep one factor per kind of work. If you do landing pages and also six-week builds, the two may not overrun the same way. Measure them separately once you have three of each. Until then, one factor is still better than none.

Ignore one outlier, once. The project where the client vanished for a month is not a data point about your estimating. Take it out, note that you did, and put the next one like it back in.

Re-measure every quarter. The factor drifts as you get better at the work and as your estimates learn from it. A factor that was 1.3 and is now 1.15 is a raise you gave yourself.

What to do with the factor before the next quote

Multiply the estimate. 30 hours at 1.3 is 39. Quote 39. You are not padding; padding is a guess added to a guess. You are replacing the plan’s number with the measured one.

Or keep the number and know the rate. Sometimes you quote 30 anyway, because the relationship is worth it or the client has a budget. Fine, as long as you do it knowing that 3,000 over 39 hours is 77 an hour and not 100, and that 77 sits under a floor of 85. A discount you did not mean to give is still a discount.

On fixed price, check the factor against the hour budget. The fee divided by your floor is the most hours the project can afford. If the factor-adjusted estimate is above that number, the fee is wrong before the project starts, and it is far cheaper to say so now than at hour 40.

On hourly, quote the range honestly. “About 30 hours, in practice 30 to 40” is a truer sentence than “30 hours”, and clients who have hired freelancers before recognise it as the sentence of someone who tracks. The hourly-versus-fixed question gets easier once you know your factor: a factor near 1.1 means your estimates are good enough to take fixed-price risk; a factor near 1.5 means bill hourly until they are.

Set the project’s budget to the hours you quoted. Not to the factor-adjusted number. The point of the budget is to see the overrun as it happens, and it happens against the quote.

Where CronLoom comes in

CronLoom does not write estimates and does not compute the factor for you. What it keeps is the two numbers the factor is made of, per project.

Give the project a budget in hours, the 30 you quoted, and log the hours against it as they happen. The project shows how much of the budget is used and, once it passes, how many hours it is over. The budget alert arrives at 80 percent, which on a 30-hour quote is hour 24, and again at 100 percent. That is the overrun becoming visible while there is still something to do about it.

At the end, the project’s tracked hours over its budget hours is that project’s factor, and its effective rate is the fee over the tracked hours, flagged if it landed under the floor you set. Three finished projects later you have the number this whole guide is about, read off your own work rather than a study of students.

The effective hourly rate calculator does the single division for one project if you want to try it by hand. If you want the budget and the flag running while the hours happen, join the waiting list.

Try it on your last three projects

Open the last three projects you finished. Write down what you quoted and what you tracked. Divide. If the three numbers are close to each other, you have just found the most useful number in your business, and the next quote you write should be multiplied by it.

If they are all over the place, that is a finding too: it means the overrun is not yet about you, it is about the projects, and the halfway check is the tool until it settles.