Completing the Square vs. Quadratic Formula: Which to Use and Why It Matters

For any quadratic equation $ax^2 + bx + c = 0$ with $a \neq 0$, the roots are

Here $a = 1$, $b = 6$ and $c = 5$, so

which gives $x = -1$ and $x = -5$. Substituting back confirms both: $(-1)^2 + 6(-1) + 5 = 0$ and $(-5)^2 + 6(-5) + 5 = 0$.

That answers the first half of the question and none of the second. The roots tell you where the curve meets the horizontal axis, not how low it goes. The turning point does sit halfway between them, at $x = -3$ — the average of the two roots is always $-\frac{b}{2a}$ — but that is an extra move you have to know to make.

Completing the square rewrites the expression instead of solving it. For any $x^2 + bx + c$,

because expanding the square gives $x^2 + bx + \frac{b^2}{4}$, and the $-\frac{b^2}{4}$ on the right cancels that extra $\frac{b^2}{4}$. For our quadratic, $\frac{b}{2} = 3$:

Read the answer off the page. A square is never negative, so $(x+3)^2 - 4$ is smallest when $(x+3)^2 = 0$. The minimum value is $-4$, it happens at $x = -3$, and the vertex is $(-3, -4)$. Checking directly, $(-3)^2 + 6(-3) + 5 = -4$.

The roots fall out of that same form. Setting $(x+3)^2 - 4 = 0$ gives $(x+3)^2 = 4$, so $x + 3 = \pm 2$ and $x = -1$ or $x = -5$. The same two roots, from a rewrite that also carried the vertex.

The methods are not rivals. Divide $ax^2 + bx + c = 0$ by $a$, move the constant across, and complete the square on what remains:

Take square roots, subtract $\frac{b}{2a}$, and the formula appears. It is the completed square done once with letters, so nobody repeats it. Using it keeps the result and throws away the working, which is fine when the result is all you need.

Take $3x^2 - 2x + 7 = 0$, where the only instruction is to find exact roots. Here $a = 3$, $b = -2$, $c = 7$:

Completing the square here is a slog through thirds. Divide by $3$ first and the work sits on $x^2 - \frac{2}{3}x + \frac{7}{3}$: halve $-\frac{2}{3}$, square it to $\frac{1}{9}$, add and subtract that, then multiply back through by $3$ to reach $3x^2 - 2x + 7 = 3\left(x - \frac{1}{3}\right)^2 + \frac{20}{3}$. The work is not wasted — it puts the vertex at $\left(\frac{1}{3}, \frac{20}{3}\right)$, well above the axis, which is why no real root exists.

That cost is not a quirk of imaginary numbers. It turns up with ordinary real roots too: for $5x^2 + 7x - 8 = 0$ you divide by $5$, halve $\frac{7}{5}$ to $\frac{7}{10}$, and square that to $\frac{49}{100}$ before anything simplifies. If only the roots were wanted, none of that arithmetic is paid back and the formula got there in one line. Ask for vertex form and the formula gives nothing back.

The Sunday-night mistake is not a gap in algebra. You can complete the square; you just did not think to. In one experiment with students new to solving linear equations, one group was asked to re-solve a finished problem using a different order of steps^[1]. Those students came out more flexible than students who worked through the same problems as usual^[1]. The nudge was needed, which suggests that, left alone, learners tend to repeat the one routine they were taught^[1].

Knowing a second method is not the same as reaching for it. In a three-week summer course for students who struggle with algebra, students came away knowing more than one method and valuing having more than one, yet did not always use the second^[2]. Asked why they chose as they did, they named how familiar a method felt as one reason^[2]. The formula is the familiar one. It is on the formula sheet and it never fails, so it gets written before the question has finished being read.

A general finding sits behind that advice. A review that pooled 57 experiments found that learners who studied two cases side by side usually learned more than learners taught other ways, including those who studied the same cases one after the other^[3]. Those experiments spanned many subjects, in laboratories and classrooms^[3].

Algebra has its own version. In one experiment, seventh-graders who studied two solution methods side by side got better at carrying out the steps, and at picking a method, than classmates who studied the same two methods one after the other^[4]. On understanding the ideas behind the steps, the two groups ended up about the same^[4]. A later experiment with seventh- and eighth-graders sharpened what is worth comparing^[5]. Students who compared two methods for one problem understood more, and chose better, than students who compared two problems solved by the same method^[5].

At your desk that means one quadratic, both routes, side by side on one page. Then write a sentence about what each route gave you that the other did not. Ten quadratics by formula and then ten by completing the square is closer to the arrangement that did less for those students^[4][5]. If a classmate, a teacher or a Learn4Less tutor works through it with you, the useful question for them to ask is not "is this right" but "what did the question ask for".

Order matters too. In another experiment with seventh- and eighth-graders, those who had already tried using algebra on the equations got more out of comparing two methods^[6]. Those who had not yet tried were better off studying the worked examples one at a time, or comparing problems of different types^[6]. So if completing the square still feels shaky, steady it on two or three simple quadratics first, then compare.

Every experiment above is about learning to solve equations, not about quadratics. Three used seventh- or eighth-graders, one a summer class for students struggling with algebra, and one students who had never been taught formal equation solving^{[1][2][4][5][6]}. A student rewriting a parabola is doing something none of them tested.

None of this work compares completing the square with the quadratic formula^{[1][2][4][5][6][3]}. So none of it says which method is faster, safer, or better for a given question. What the algebra above does settle, proved on the page and needing no study behind it, is narrower and enough: both routes reach the same roots, and only one of them ends holding the vertex.

One more limit. In the pooled review, the advantage of comparing cases was clearest when learners were tested straight afterwards^[3]. So comparison is well supported as a way to understand two methods today, and much less well supported as a promise the distinction stays automatic months later^[3].

The decision takes five seconds, before any writing. If the question wants roots, the formula is enough, and it is the shorter route when the numbers are ugly. If it wants the vertex, a maximum or minimum, or vertex form, complete the square, and collect the roots on the way.