Humans Are Terrible Estimators – How To Make Better Time Estimates

Reading time: 2 min

How often do you underestimate the time required to complete something?

Our ability to make time estimations is terrible.

From how much time we require to complete a task, to how much energy we consume, we tend to be way off from reality.

One study asked students to estimate how long it would take them to finish a paper under:

1. realistic conditions (the best realistic estimate),
2. perfect conditions (“everything went as well as it possibly could”), and
3. worst possible conditions (“everything went as poorly as it possibly could”).

Their realistic condition estimates are much closer to “perfect condition” estimates than their “worst possible condition” estimates.

Moreover, not even half of the students finished their papers under their worst possible condition time estimates.

See the original study here, by Buehler, R., Griffin, D., & Ross, M. (1994).

What can we do about this, and how can we use this to our advantage?

Well, limiting time allocation in estimates does have benefits; this leverages Parkinson's Law (work expands to fill the time allocated). We tend to be more efficient when given less time to complete a task.

What we can do is:

1. Estimate with the worst possible conditions (given that this is the closest to the real amount of time required).
2. Add a (generous) buffer time before the next event or task.

The buffer time will prevent us from overrunning into the next task or event, and gives a reward for finishing early (rest or no-guilt leisure time).

Getting accurate time predictions:

Interestingly, the same study notes that we're much better at estimating how long it would take someone else to complete a task than if we were to complete it ourselves.

If we require an accurate estimation, we can ask others to provide a time estimate for us.

Alternatively, we can start marking down our time requirement predictions and actual time requirements down. Divide the actual time requirement by our prediction to get a “prediction to reality ratio” to multiply our future predictions with.

For example, if I estimate that this post will take me 1 hour to write but took 1.5 hours in reality, my “prediction to reality ratio” is:

Hours in Reality / Hours in Prediction = 1.5/1 = 1.5

I should multiply my future time estimations with this ratio to get a more accurate estimate. Get an average of multiple ratios for even better results.

Related:

• Broken snooze sleep is worse than no sleep
• Facing Overcommitment / Overburdening - Forecasts are Inaccurate