What Big-Oh Measures

First, the timing measures the whole system, not just the algorithm. Change the machine, the language, the compiler, or what else is running at the time, and the number changes, even though the algorithm did not.

Second, the timing is for one input size only. The question is how the work grows as the input grows, and a single run at one size does not answer that question.

The coefficients reflect the expression of the algorithm (the way the code is written), and in general the machine, the language, and the compiler. The method is $O(n)$.

It says the work grows in proportion to the input size: double the input size and the work roughly doubles. It leaves out the exact step count: the number of instructions it carries out to search an array of size $n$.