Fermi Estimator

decompose any unknown into multiplicative factors w/ low/mid/high - get a 90% range
How: name what you want to estimate (market size, # of piano tuners in chicago, eyeballs your pitch will reach). Break it into multiplicative factors. For each, give a low / mid / high guess (90% confidence range). The tool runs 10,000 monte carlo trials sampling each factor lognormally, returns the distribution of products. The geometric mean is your point estimate; the 5th-95th percentile is your honest range.
Example: "monthly revenue from this side project" = (visitors/mo) × (signup conv) × (trial-to-paid) × ($/user). Your gut says low/mid/high for each, the tool tells you whether the spread is meaningful or whether you've fooled yourself with false precision.
What Are You Estimating?
Unit
Factors (multiplicative)
Estimate
distribution of 10,000 monte carlo trials (log scale)
Archive
client-side - localStorage - lognormal sampling between low & high (90% CI), product across factors
"better roughly right than precisely wrong" - keynes (also fermi liked this one)