Unlike adding or subtracting numbers, you can't just multiply any two matrices together. They have to be compatible, following a strict rule. Think of it like a handshake. For two matrices to connect, their “inner” dimensions must match.

Let's say we have Matrix A with dimensions $m \times n$ (that's $m$ rows and $n$ columns) and Matrix B with dimensions $p \times q$. To multiply them in the order $A \times B$, the number of columns in A ($n$) must be exactly the same as the number of rows in B ($p$).

We can write their dimensions side-by-side to check compatibility:

The dimensions $n$ and $p$ are the inner dimensions. If they match, the matrices are considered conformable for multiplication. If they don't, the operation is impossible. It's a simple, non-negotiable check.

If the inner dimensions match, the multiplication is valid. But what size will the new matrix be? That's where the outer dimensions, $m$ and $q$, come in. The resulting matrix will always have the number of rows from the first matrix and the number of columns from the second.

So, multiplying an $m \times n$ matrix by an $n \times p$ matrix gives you a brand new $m \times p$ matrix. The inner dimensions ($n$) disappear, and the outer ones form the shape of the product.

Notice that order matters. If we try to multiply the matrices from the first example in reverse order ($B \times A$), we'd be checking a $3 \times 4$ against a $2 \times 3$. The inner dimensions would be 4 and 2. Since they don't match, the multiplication $B \times A$ is undefined, even though $A \times B$ worked perfectly.

This is a huge difference from regular multiplication, where $5 \times 4$ is the same as $4 \times 5$. With matrices, changing the order can change everything, a property first explored in detail by the mathematician Arthur Cayley in the 1850s.

Let's check one more. Can you multiply a $4 \times 2$ matrix by a $2 \times 3$ matrix?

Before you ever calculate a single number in a matrix multiplication problem, always check the dimensions. It's a fast way to see if the problem is even solvable.