Numbers without a data team
Maya reads your finding and asks one question: how many people does this actually hit? The data team can tell you. Next quarter.
Estimate out loud, so people argue with the guess
How Enrico Fermi handled it
Enrico Fermi
1945
At the first atomic test, Fermi wanted a rough idea of the blast's strength long before the instruments could be read.
He dropped scraps of paper as the shock wave passed, paced out how far they moved, and estimated about ten kilotons. The measured figure, which came much later, was around twenty. Off by half, right about the size of the thing, from paper and a few seconds.
An estimate with its workings showing is useful long before the precise number exists. Roughly right this week beats exactly right next quarter.
Doing it with AI, and where it breaks
Have a model lay the estimate out as a chain, then check its arithmetic and, more importantly, every number it assumed.
Its guesses come from the average of the internet: US prices, US conversion rates, US salaries. Give it Sona's numbers, or make it mark every one it made up.
I need to estimate [the number, with a unit and a time period]. Break it into no more than five multiplied factors. For each, give a low and a high guess and say exactly where the guess comes from. Mark any factor you made up. Then tell me which factor the result is most sensitive to.
Build it
A sized opportunity: five guesses or fewer, each with a reason, and a low, middle and high.
Google Sheets
The worked solution
Stuck, or want to compare?
There is a worked solution. Try the build first. Reading it before you have attempted anything is the fastest way to learn nothing.
Checkpoint
If you did the build, these take two minutes. If you cannot answer one of them, that is the part to go back to.
Moving on marks this lesson complete. Finish the build first, it is the part that counts.