Verystep Guides Verystep Guides
How to Multiply Matrices and Check the Result
Matrix multiplication has one gate to pass before any arithmetic β and one hand check afterwards that catches almost every wrong grid.

Multiplying matrices isn't multiplying numbers entry by entry. Each entry in the product comes from a whole row of the first matrix paired with a whole column of the second: multiply the pairs, add them up, and that sum takes one seat in the result. Before any of that, the shapes have to fit β the columns of the first matrix must match the rows of the second, or there is no product at all. This guide runs one full example through the calculator, then shows you the two checks that matter: the shape check before you calculate, and recomputing a single entry by hand after.
The one rule that decides whether a product exists
Write the shapes down first, always in the same order: rows Γ columns. A matrix with 2 rows and 3 columns is 2Γ3. Now the rule. You can multiply A Γ B only when the number of columns in A equals the number of rows in B. The inside numbers of the two shapes have to match. If they don't, stop. There is no product, no partial answer, and no trick that rescues the pairing β the rows of A would run out of partners in the columns of B, or the other way round.
When the inside numbers do match, they also tell you the shape of the answer. The product takes the outside numbers: A rows Γ B columns. So a 2Γ3 times a 3Γ2 gives a 2Γ2 product. The 3s agree in the middle and drop away, and the 2 and the 2 on the ends become the result's shape. Say it out loud before you calculate. If the grid that comes back isn't that shape, something went wrong with the setup, not with the arithmetic.
Multiply two matrices in seven steps
Open the Matrix Calculator and work down this list. The multiplication is the middle step β the steps around it are what make the result trustworthy.
- Enter matrix A and matrix B with their rows and columns set first, then fill each entry in order and double-check the shapes shown before you multiply.
- Check compatibility before you calculate: the number of columns in A must equal the number of rows in B, or the multiplication cannot be done.
- Read the product shape from the outside numbers β rows of A by columns of B β and confirm the result grid matches that shape before you trust any entry.
- Multiply to get the product, then pick one entry and recompute it by hand as a row of A dotted with a column of B, adding each paired product in order.
- Compare your hand-checked entry with the same position in the result, and if it differs, re-enter that row and column rather than changing the whole matrices.
- Try the reverse order only as a separate run, and expect a different shape or different entries β do not assume A times B and B times A will match.
- Use the other operations β add, subtract, determinant, inverse, and transpose β in separate runs, keeping the multiplication result you already checked unchanged.
Notice where the hand check sits: after the full product, on one entry only. You're not redoing the tool's work. You're sampling it. One entry recomputed honestly, in the right row and column, catches transposed entries, swapped matrices, and most typing slips β the errors that actually happen.
A worked product: 2Γ3 times 3Γ2
Take A as the 2Γ3 matrix [[2, 3, 1], [4, 1, 5]] and B as the 3Γ2 matrix [[6, 2], [1, 3], [5, 4]]. Compatibility first: A has 3 columns, B has 3 rows, so the product exists, and its shape is 2Γ2. Enter both, multiply, and the calculator returns [[20, 17], [50, 31]]. Hold that grid up against the shape you predicted before you look at any single number. Two rows, two columns. Good. Now the entries earn their turn.

Each entry is a row of A dotted with a column of B. Top-left: row one of A, [2, 3, 1], against column one of B, [6, 1, 5] β 2Γ6 + 3Γ1 + 1Γ5 = 20. Top-right: row one against column two, [2, 3, 4] β 2Γ2 + 3Γ3 + 1Γ4 = 17. Bottom-left: row two of A, [4, 1, 5], against column one β 4Γ6 + 1Γ1 + 5Γ5 = 50. Bottom-right: row two against column two β 4Γ2 + 1Γ3 + 5Γ4 = 31. Four entries, four short sums, and the grid [[20, 17], [50, 31]] is accounted for in full.
Checking the result without redoing everything
On a bigger pair you won't recompute all four β or nine, or sixteen β entries by hand. Pick one, and pick it well. The bottom-right entry is a good default, because it uses the last row and the last column, where tired typing tends to land. Recompute it exactly as above: walk across the row and down the column at the same pace, multiply each pair, and keep a running total. If your sum matches the grid, the setup was almost certainly right, because a wrong shape or a swapped entry rarely survives even one honest dot product.
If it doesn't match, resist the urge to re-enter everything. The mismatch lives in one row or one column β the two lines you just used. Compare those entries against your source, fix the one that's off, and multiply again. Then check a different entry this time, so the new run gets its own sample. Two clean spot checks on separate entries, plus the shape check up front, is a solid standard for trusting a product you didn't compute by hand.
Order matters, and other traps
Ordinary multiplication lets you swap the factors. Matrix multiplication doesn't. A Γ B and B Γ A are different questions, and here they don't even have the same shape: A Γ B is 2Γ3 times 3Γ2, giving 2Γ2, while B Γ A is 3Γ2 times 2Γ3, giving a 3Γ3 product. Both exist in this case, but they are different grids with different entries. In other pairings the reverse won't exist at all. So never flip the order to save typing, and when a problem states an order, enter the matrices in exactly that order.
Two more traps. First, don't multiply entry by entry β pairing the top-left of A with only the top-left of B gives you a same-shaped grid that looks plausible and means nothing here; that is a different operation, not this one. Second, watch rows versus columns when you set the shapes. Entering A as 3Γ2 instead of 2Γ3 doesn't produce a small error. It produces a different matrix, and every entry after it is wrong in a way no spot check on the real problem can fix. Shapes first, entries second, multiply third β in that order, every time.
The Verystep Matrix Calculator multiplies two matrices once you set their shapes and entries, and it also handles add, subtract, determinant, inverse, and transpose in separate runs. Enter the 2Γ3 and 3Γ2 example from this guide first, check you get the same 2Γ2 product, then run your own pair.
Last verified November 16, 2026 β steps and worked examples checked against the Verystep Matrix Calculator and verified by hand calculation.
Frequently asked questions
Why did the calculator refuse to multiply my matrices?
Almost certainly the shapes don't fit. Multiplication needs the columns of the first matrix to equal the rows of the second β a 2Γ3 can multiply a 3Γ2, but not a 2Γ2, because the inside numbers, 3 and 2, don't match. Check the shape you set for each matrix first. A matrix entered with its rows and columns swapped is the most common cause.
Is A times B ever the same as B times A?
It can happen for special pairs, but you can't count on it, and usually it isn't. In this guide's example the two orders don't even share a shape: A Γ B is 2Γ2, while B Γ A is 3Γ3. Treat the reverse order as a new problem every time β check its compatibility and shape from scratch, and run it as its own calculation.
How many entries should I check by hand?
One is the minimum, two is better on anything you'll rely on. Pick entries from different rows and columns β the bottom-right plus one other, say β and recompute each as a row dotted with a column. A single match already rules out the big setup errors. Two matches on separate lines make a wrong grid very unlikely to slip through.
What does the calculator do besides multiplication?
It also does add, subtract, determinant, inverse, and transpose. Run each as its own operation, with its own setup checks β addition needs both matrices at the same shape, for instance, which is a different rule from multiplication. This guide covers multiplication and its checks; the habit of verifying the shape before and sampling an entry after carries over to the rest.
My hand check is off by a little. Did I pick the wrong row or column?
That's the first thing to suspect. An entry in row two, column one is built from row two of A and column one of B β mixing in column two instead is an easy slip, and it usually produces a near miss rather than a wild number. Redo the dot product slowly, saying each pair out loud. If it still disagrees, compare that row and column against your source numbers.
More guides

How to Use an Auto Loan Calculator to Compare Two Offers
Two dealers can quote payments that look almost identical and still be selling you loans that cost hundreds of dollars apart. The calculator shows you where the gap hides.

How to Build a Debt Payoff Order: Avalanche vs Snowball, Step by Step
Paying extra money toward debt only works if the extra money has a target. This guide builds that target: an ordered list, made two different ways, so you can see exactly where each spare dollar goes.

How to Project Compound Growth (and Read the CAGR Mode)
This calculator does two jobs in one place: it projects a balance forward at a rate you set, and it works backwards from two balances to the steady rate that links them.