Your Jobs Don't Run Long by Accident. Your Estimates Are Built to Miss.

Think about the last restoration job that finished early. Not on time. Early. Most owners can't name one. The dry-out took an extra two days, the rebuild waited on a cabinet order, the homeowner changed the flooring halfway through. Each overrun had its own reason, and each reason sounded like bad luck.

It isn't bad luck when it happens every time. It has a name, it has been measured, and you can price for it.

The idea: the planning fallacy

Daniel Kahneman and Amos Tversky coined the term in 1979 for a specific habit: holding a confident belief that your own project will go as planned, even while knowing that most similar projects ran late.

In 1994, Roger Buehler, Dale Griffin and Michael Ross put numbers on it in the Journal of Personality and Social Psychology. Their first study was deliberately hard on the effect. They phoned 37 honors thesis students at the University of Waterloo in their final semester, when the project was important, well underway and close to done, and asked each one when they would hand it in. Then the course coordinator recorded the date each thesis actually arrived.

The results, for the 33 students whose finish dates were recorded:

• Their best estimate averaged 33.9 days. They actually took 55.5 days.
• Only 29.7% finished by the date they called their most accurate prediction.
• Asked for a date "if everything went as poorly as it possibly could," they averaged 48.6 days. Fewer than half, 48.7%, finished even by that worst case.

Two more findings from the same paper matter more for a business owner than the headline.

First, the students knew better. In a classroom survey, students reported finishing about two thirds of their previous projects later than they expected. They had the history. They just didn't use it. The researchers found people build a story about how this task will go and treat past overruns as special cases that don't apply this time.

Second, the estimates weren't useless. Predicted and actual times were strongly correlated. Students who expected to take longer did take longer. The estimates had real information in them. They were just shifted, consistently, in one direction.

The authors also make an observation that should sound familiar: we are not surprised when a colleague's house renovation takes twice the time they predicted. We see other people's overruns coming. We don't see our own.

The restoration translation

A restoration estimate is exactly the kind of forecast this research describes. You walk the loss, you picture the job going right, and you write down the hours and days that story requires. Then the job happens in the real world, where drying takes what it takes, materials arrive when they arrive, and scope is discovered behind the drywall rather than seen on the walkthrough.

That costs you in three places at once.

Labor. If the estimate assumed the story and the job delivered the history, the extra crew hours come straight out of margin. On an insurance job the price is set by the approved scope, not by how long you took, so a slow job doesn't bill more. It just earns less.

Capacity. The crew still on Monday's job isn't starting Tuesday's. Every overrun pushes the next job, which is how a shop can feel slammed and still come up short on revenue for the month.

Cash. This is the one nobody connects. On a typical claim, final billing and whatever the carrier is holding back wait on completion. A job that runs 60% long doesn't just cost more labor. It pushes back the day you can finish billing it, while payroll for all those extra hours went out every Friday. A planning miss is a cash timing miss.

And the worst-case finding closes the obvious escape hatch. "I'll just pad it" is the move most owners make, and the students who were told to picture everything going wrong still undershot. A number you make up to feel conservative is still a number you made up.

Five moves

1. Build your own reference class. Pull your last 20 or 30 closed jobs and put estimated hours next to actual hours, and estimated days next to actual days. Split mitigation from reconstruction. That ratio is the most valuable number in your estimating process, and almost nobody has it.
2. Adjust the estimate, don't replace it. The research says your estimates carry real information and a consistent bias. So keep estimating the way you do, then apply your historical ratio. If your reconstruction jobs run 1.4 times their estimated duration, schedule and forecast at 1.4.
3. Get a second estimator on duration. People forecast other people's projects more realistically than their own. Have someone who didn't walk the loss look at the scope and say how long it will take. Where the two numbers disagree is where the story is doing the work.
4. Stop trusting the worst case. A gut "worst case" is still a story. Your closed-job history isn't. Use the history for contingency, and write down what caused each overrun so the next estimate can see it coming.
5. Forecast cash on actual duration, not estimated duration. If the job will really finish in week seven, the final invoice goes out in week seven, and your cash forecast should say so. Build the 13-week forecast off the adjusted timeline, not the one on the estimate.

None of this requires estimating better in the moment. It requires admitting what your own data already says and letting it correct the number before you commit crews and cash to it.

Turning closed-job history into an estimating ratio and a cash forecast is exactly the work of the CFO seat in a restoration company. If you want to see what your own jobs say about your estimates, let's talk.

General education for restoration and reconstruction owners, not financial, accounting or legal advice for your specific business. Talk to your own advisors before changing how you price or schedule work.

← Back to all posts