Anchor on what you know and scale from there

Start from a number you are confident about, then reason to the unknown.

Why it works

Known anchors reduce the range of uncertainty by giving the estimate a grounded starting point. Even rough anchors — the population of a city, the price of a common item, your own daily behaviors — constrain the estimate more than no anchor at all. Anchoring on knowns before inferring unknowns is the structural move that gives Fermi estimates their surprising accuracy despite limited inputs.

How to do it

1. Before estimating the unknown, identify what you do know that is related: a count, a rate, a size.
2. Scale from that anchor using ratios or multipliers you can justify.
3. Use your own experiences as anchors where possible — they are more reliably known than general statistics.
4. Check the anchor against a second independent anchor if one is available.

Evidence

Reference point reasoning is consistent with how expert estimators operate. Anchoring on known quantities and adjusting is more accurate than estimating in a vacuum, though adjustment from anchors is often insufficient (the anchoring bias applies here too). (mechanistic)

The anchoring bias means people under-adjust from their initial anchor; using multiple independent anchors and averaging provides a correction.

Common mistake

Using an anchor that is itself highly uncertain or easily manipulable, then adjusting insufficiently from it — inheriting the anchor’s error rather than correcting for it.