48 × 23 Estimates Badly and 52 × 19 Estimates Well
Published 11/4/2026 · 4 min read · Business tools
Rounding both factors before multiplying is the standard way to sanity-check a product, and its accuracy depends entirely on the directions of the two roundings. In 48 × 23, rounding 48 up to 50 adds 4.2 % and rounding 23 down to 20 removes 13 % — both errors act on the product in the same direction, and the estimate of 1 000 falls 9.4 % short of the true 1 104. In 52 × 19, rounding 52 down to 50 removes 3.8 % while rounding 19 up to 20 adds 5.3 %, and the two nearly cancel: the same estimate of 1 000 is within 1.2 % of the true 988.
Both round to 50 × 20 = 1 000. The first is 9.4 % off and the second 1.2 %, because in one case the two roundings pull the same way and in the other they cancel.
Round in opposite directions on purpose
The practical rule that falls out of this is worth adopting: when estimating a product, round one factor up and the other down. It costs nothing, it makes the arithmetic no harder, and it converts a potential 10 % error into a likely 1 % one. Division reverses the logic — rounding the numerator up and the denominator up both push the answer in opposite directions, so for a quotient you should round both the same way. Getting this backwards is what makes 950 ÷ 31 estimated as 1 000 ÷ 30 come out 8.8 % high, when rounding both down to 900 ÷ 30 would have given 30 against a true 30.6.
An estimate is a claim about the digits you keep
Estimating to one significant figure claims the answer's leading digit and nothing more, which is exactly right when checking whether a spreadsheet has misplaced a decimal point and useless for checking whether a quote is competitive. The 48 × 23 case shows the limit: 1 000 and 1 104 share no digits at all, so a one-figure estimate cannot even confirm the leading 1. Two figures — 48 × 23 as 48 × 23 rounded to 50 × 23, giving 1 150 — costs one more multiplication and lands within 4 %. Choosing how many figures to keep is the whole of the skill, and keeping one is usually a decision about how much you care rather than how quickly you must answer.
| Calculation | Estimate and truth | Error |
|---|---|---|
| 48 × 23 | 1 000 vs 1 104 | 9.4 % |
| 52 × 19 | 1 000 vs 988 | 1.2 % |
| 297 × 41 | 12 000 vs 12 177 | 1.5 % |
| 950 ÷ 31 | 33.3 vs 30.6 | 8.8 % |
Worked with our own calculator
Estimation calculator (rounding)
Given
- First number
- 24
- Second number
- 12
- Operation
- Add (+)
Result
- Estimate
- 30
- Exact answer
- 36
- Error
- 16.67%
These figures are produced by the calculator below, not typed in by hand — they are recomputed whenever the tool changes.
Run it on your own figures →Frequently asked questions
- Does the same logic apply to sums?
- It applies more gently, because addition does not compound errors the way multiplication does. Rounding 48 and 23 to 50 and 20 gives 70 against a true 71 — an error of 1.4 %, where the same roundings in a product gave 9.4 %. That is because in a sum the absolute errors add, while in a product the relative errors add. So estimating a long column of figures by rounding each is far safer than estimating a chain of multiplications the same way, and the sum of many roundings in mixed directions tends to cancel out further still.
- How wrong can a two-factor estimate get?
- Roughly the sum of the two relative roundings, and worse when the numbers are small. Rounding 4 to 5 and 3 to 5 is a 25 % and a 67 % error, giving 25 against a true 12 — more than double. The lesson is that the technique needs numbers with enough digits for the leading one to be stable: rounding to one significant figure is reliable in the hundreds and reckless in the units. Whenever the factors are under about twenty, exact arithmetic is fast enough that estimating buys nothing and risks a great deal.
Articles you may find interesting
All guides →Related tools
Sources
Spotted a mistake in this article?