Here's another way to tell the story, with an example, if it helps.
If you want to solve an equation like
[tex]
3 x^2 + 2 x - 5 = 0[/tex]
first, you divide the whole thing by 3 (in order to leave the [tex]x^2[/tex] alone), so that now it looks like
[tex]
x^2 + \frac 2 3 x - \frac 5 3 = 0[/tex]
Now, you take
half the coefficient of x, that is, the half of [tex]\frac 2 3[/tex] which is [tex]\frac 1 3[/tex]
and "complete the square" by building this (explanation later):
[tex]
\left(x + \frac 1 3 \right)^2 - \left( \frac 1 3 \right)^2 - \frac 5 3 = 0[/tex]
The [tex]- \left( \frac 1 3 \right)^2 - \frac 5 3[/tex] part is a constant, that you can combine into a single number, which in this case is [tex]-\frac 1 9 - \frac 5 3 = \frac {-1 - 15} 9 = -\frac {16} 9[/tex]
so that your equation now looks like
[tex]
\left(x + \frac 1 3 \right)^2 - \frac {16} 9 = 0[/tex]
and since
x appears only once, now you can solve for it:
[tex]
x = \pm \sqrt {\frac {16} 9} - \frac 1 3 = \pm \frac 4 3 - \frac 1 3[/tex]
giving the two solutions, 1 and -5/3.
Easy!
. . .
The trick was to change this:
[tex]
x^2 + \frac 2 3 x[/tex]
into this:
[tex]
\left(x + \frac 1 3 \right)^2 - \left( \frac 1 3 \right)^2[/tex]
If you expand [tex]\left(x + \frac 1 3 \right)^2[/tex] you will notice why.