TL;DR Architects underestimate for 4 reasons: planning fallacy, professional optimism, anchoring on a model project and the desire to close the contract. The solution is not "being more realistic", it is changing the process: range-based estimates, a multiplier factor, a weekly checkpoint.

The 4 biases

1. Planning fallacy

A universal bias: we estimate based on the best-case scenario, ignoring problems. Documented by Kahneman and Tversky since 1979, people in general are off by 30 to 70% on the low side.

2. Professional optimism

"I am good at what I do." That is true. But "being good" does not speed up an indecisive client, a late supplier, frozen software.

3. Anchoring on a model project

"This one is similar to the one I did last month, I spent 90h." But you remember the useful hours; you forgot the revisions and rework.

4. The desire to close

Underestimating time keeps the price attractive. Unconsciously, you "adjust" the estimate to fit the budget the client mentioned.

Practical consequences

  • Margin drops 15-35% on typical projects.
  • Team overloaded, quality slipping.
  • The next estimate is based on a previous wrong estimate, so the bias compounds.

5 antidotes

  1. Range-based estimate (e.g. "120-180h" instead of "150h"). It acknowledges uncertainty.
  2. Multiplier factor of 1.4× applied on top of your intuitive estimate.
  3. Mandatory tracking for the first 6 months. Without real data, the bias persists.
  4. Compare with an external benchmark, do not rely only on your memory.
  5. Buffer of 10-15% in the schedule for revisions and unforeseen issues.
Limify

Estimates calibrated with your history

Limify shows the real hours of your past projects when it is time to quote the next one. Cognitive bias tackled with data, not with willpower.

Try it free