Think about what a spreadsheet actually is: a grid of cells, each holding something — a number, a word, a date, or nothing at all. The moment you have a few hundred such cells, two questions become unavoidable. How many of these cells actually contain something? And when a calculation spits out ₹1,247.8364, how do I present it as money?
Statistical functions answer the first question. Rounding functions answer the second. They look unrelated, but they share a purpose: turning a raw pile of data into something a human can read and trust.
Counting: the question behind the function
Suppose you have a column of daily sales figures, and a few days were holidays so those cells are blank. You want to know how many days actually had sales.
COUNT does exactly this. It walks through the range you give it and counts only the cells that hold numbers. Blank cells are skipped. Text is skipped. A cell containing the word "N/A" is skipped, because it is text, not a number.
COUNTA counts differently. The "A" stands for "all" — it counts every cell that is not empty. Numbers, text, dates, error values, even a single space typed into a cell: all of these count. Only genuinely blank cells are ignored.
That single distinction is the whole idea:
| Function | Counts | Ignores |
|---|
COUNT | Cells containing numbers | Blanks, text, logical values |
COUNTA | All non-empty cells | Only truly blank cells |
So if your sales column has 30 rows, 25 with numbers and 5 blank, COUNT returns 25 and COUNTA returns 25 as well — because the blanks are empty. But if those 5 cells instead contained the text "holiday", COUNT still returns 25 while COUNTA returns 30.
A quick way to remember: COUNT is fussy — it only wants numbers. COUNTA is generous — anything that isn't empty will do.
This matters in practice. If you use COUNTA to count entries and then divide a total by it to get an average, and some of those entries are text labels, your average will be wrong. COUNT protects you from that mistake when you genuinely want the number of numeric data points.
Rounding: why the raw answer is not the final answer
Now the second problem. You compute a total, or a tax, or a per-unit price, and the machine gives you something like 1247.8364. Money is not quoted to four decimal places. You need 1247.84.
ROUND takes two arguments: the number, and how many decimal places you want.
ROUND(1247.8364, 2)=1247.84
The rule it follows is the one you learned in school: look at the digit immediately after the place you are keeping. If it is 5 or more, round up; otherwise round down. Here the third decimal is 6, so the second decimal goes from 3 to 4.
The second argument is more flexible than it first appears. It can be zero, to give a whole number. It can even be negative, which rounds to the left of the decimal point:
ROUND(1247.8364, 0)=1248
ROUND(1247.8364, −2)=1200
A negative second argument rounds to the nearest ten, hundred, thousand, and so on. Useful when you are estimating at scale.
Rounding is not the same as displaying fewer decimals. If a cell shows 1247.84 because you narrowed the column or applied a number format, the cell still holds 1247.8364 underneath, and every later calculation uses the full value. ROUND actually changes the stored number. When your computed amounts must match what appears on an invoice, you want ROUND, not formatting.
The family of rounding helpers
ROUND is the balanced one — it goes to the nearest value. Two relatives always push in one direction:
ROUNDUP always moves away from zero, no matter how small the discarded part is. ROUNDUP(1247.831, 2)=1247.84.ROUNDDOWN always moves toward zero, discarding the remainder. ROUNDDOWN(1247.839, 2)=1247.83. …