Mental math for estimating

You do not need to be fast at arithmetic to estimate well. You need about five tricks, and all of them are about getting rid of work rather than doing it.

Zoggle · a guide for people who like numbers

The mental arithmetic that estimation requires is not the arithmetic you were drilled on at school. Nobody needs the exact product of 847 and 1,932. What you need is to know within a second that it’s “about 1.6 million,” and to be sure about the number of zeros.

Separate the digits from the zeros

This is the foundational habit and it makes everything else easier. Split every number into a small leading figure and a power of ten:

Multiply the small numbers, add the exponents, done. The mistakes that ruin estimates are almost never in the 8 × 2; they are in losing track of how many zeros were involved. Handling the zeros separately, as addition, makes them hard to lose.

If exponent notation feels awkward, use words instead — “eight hundred times two thousand is sixteen hundred thousand” works identically. The point is to keep magnitude and digits in separate mental buckets.

Round in opposite directions

When you round two numbers you are about to multiply, push one up and the other down. Rounding 47 × 62 to 50 × 60 gives 3,000 against a true 2,914 — the errors partly cancel because one number grew and the other shrank.

Round both the same way and the error compounds: 50 × 70 would give 3,500, wrong by 20%. The habit costs nothing and roughly halves your typical error.

Useful equivalences

• × 5 is ÷ 2 then × 10. 86 × 5 → 43 → 430.
• ÷ 5 is × 2 then ÷ 10. 340 ÷ 5 → 680 → 68.
• × 25 is × 100 ÷ 4. 32 × 25 → 3,200 ÷ 4 → 800.
• × 9 is × 10 minus itself. 47 × 9 → 470 − 47 → 423.
• 15% is 10% plus half of it. 15% of 80 → 8 + 4 → 12.

None of these are impressive individually. Together they eliminate most of the arithmetic that would otherwise make you reach for a calculator mid-thought and lose the thread of the estimate.

The rule of 72

Divide 72 by a growth rate to get the doubling time. Something growing 6% a year doubles in about twelve years; something growing 3% doubles in twenty-four.

This is the most useful single trick on this page, because compound growth is where intuition fails hardest. It runs backwards too: if a quantity doubled in seven years, it grew at roughly 10% annually. That reframe turns vague claims about growth into something you can check.

Squares and roots by bracketing

You rarely need an exact square root. You need to know it’s between two numbers you do know. The square root of 2,000 is between 40 (1,600) and 50 (2,500), closer to 45 (2,025) — so about 45, and that’s plenty.

Worth having permanently available: √2 ≈ 1.4, √10 ≈ 3.2, and the squares up to 25. Everything else you can bracket in a couple of seconds.

Percentages by flipping

x% of y equals y% of x, which sounds like a curiosity and is occasionally a rescue. 4% of 75 is awkward; 75% of 4 is three. Same answer, one is instant.

Checking as you go

Two habits catch most errors before they matter:

Track the units. Write them alongside the numbers and cancel them like algebra. If you wanted litres per day and your units resolve to litres per second, the arithmetic was fine and the setup was wrong.

Feel the magnitude before you calculate. Decide roughly what size answer you expect, then compute. When the result arrives three orders of magnitude away from your expectation, one of the two is wrong — and either way, you’ve caught something you would otherwise have shipped.