A small business owner once showed me her spreadsheet, certain that her bookkeeper had made a mistake somewhere. Every individual line item looked correct. Every formula checked out. But the grand total at the bottom was off by a few cents from what her bank statement showed, every single month, in a way that never quite went away no matter how many times she recalculated it. Nobody had made a mistake. The spreadsheet was rounding numbers in the background the entire time, and a few cents of rounding, repeated across enough rows, adds up to a real discrepancy.
Rounding Happens More Often Than You'd Expect
Spreadsheets generally display numbers rounded to two decimal places for currency, because nobody wants to look at a column of dollar amounts with eight digits after the decimal point. What often goes unnoticed is that the underlying value being stored and calculated with is not always the same as the rounded number being displayed on screen.
If a formula divides a total into thirds, for example, the actual stored result is a number with digits repeating forever, not a clean two-decimal figure. The cell might display $33.33, but the value the spreadsheet is actually using behind the scenes carries more precision than that. Multiply that small, invisible sliver by enough rows, and the visible totals at the bottom of the sheet can drift away from what you'd get by simply adding up the rounded numbers you can actually see.
Where This Becomes a Real Problem
For a casual personal budget, a few cents of drift rarely matters. It becomes a real problem in a handful of specific situations.
Splitting costs between people or departments. Dividing a shared bill three ways using rounded numbers can leave a cent or two unaccounted for, and reconciling exactly who owes that missing cent becomes an oddly tedious argument over essentially nothing.
Invoicing and reconciling with a bank statement. A spreadsheet total that is even a few cents off from what actually hit the bank can trigger a frustrating hunt for an error that does not actually exist anywhere in the data.
Anything calculated in stages. If one formula rounds a result and a second formula then uses that already-rounded number for a further calculation, the rounding error compounds at each stage rather than canceling out, and the final number can drift further from the truly exact value than a single rounding step ever would.
- What a spreadsheet displays and what it actually stores and calculates with are not always the same number.
- Dividing totals unevenly, like splitting a bill three ways, almost always produces a small rounding remainder.
- Rounding errors compound when one formula's rounded output feeds directly into another formula.
- The fix is usually to round only at the very last step a person actually sees, not at every intermediate stage.
Why Fractions Avoid the Whole Problem
This is really a problem with decimals specifically, not with math in general. A fraction like one third has no clean decimal equivalent, since 0.333... continues forever, but it has a perfectly exact fraction representation that never needs rounding at all. The moment you convert a clean fraction into a rounded decimal, you have introduced a tiny gap between the exact value and the number you are now working with, and that gap is exactly where the drift comes from.
The numbers weren't wrong. They were exact the whole time, right up until something forced them into a rounded decimal box that couldn't quite hold them.
Working a calculation through as a fraction, rather than converting to a rounded decimal partway through, keeps everything exact until the very last step, where you can round once, deliberately, for the final number a person actually needs to see. That single, intentional rounding step is very different from a spreadsheet quietly rounding at every formula along the way without telling you.
A Small Worked Example
Take a hundred dollar bill split three ways. Divided as a decimal, each person owes $33.33, and three people paying $33.33 each adds up to $99.99, a single cent short of the actual hundred dollars. Nobody did anything wrong. The exact answer is one third of a hundred dollars, a fraction that simply does not convert into a clean two-decimal amount no matter how you round it.
Working it as a fraction first shows you the exact value, one hundred over three, before any rounding happens at all. From there, you can decide deliberately how to handle that final cent, whether that means one person pays an extra cent, the amount gets rounded up evenly for everyone, or it simply gets written off as immaterial. The point is that the decision happens on purpose, at the end, instead of getting buried inside a formula that rounded three separate times before anyone even saw a number.
How to Stop Chasing Phantom Discrepancies
Next time a total is off by a small, stubborn amount that no individual line item seems to explain, try this before assuming someone made an error:
- Check whether the discrepancy is a few cents per row, multiplied across many rows, rather than one large error hiding somewhere.
- Look for formulas that feed rounded results into other formulas, since that is where rounding tends to compound rather than cancel out.
- Round only at the final, visible step whenever possible, rather than at every intermediate calculation along the way.
- Use exact fractions for the parts of a calculation that genuinely don't divide evenly, instead of converting to a rounded decimal earlier than you need to.
- Accept that a one or two cent gap on an uneven split is often mathematically unavoidable, not a sign anyone made a mistake.
The spreadsheet was not broken and the bookkeeper had not made an error. The numbers had simply been asked to be perfectly clean decimals when the math underneath them never actually divided that cleanly in the first place. Once you know where rounding sneaks in, those few stubborn cents stop being a mystery worth an afternoon of investigation.