The routine

A ten minute daily math facts routine that doesn’t end in tears

Written 10 September 2026 by Daniel, who built this

Here it is: same time every day, one operation, ten minutes or less, and you stop while it is still going well rather than when the questions run out. The thing worth watching isn’t how many your kid got right. It’s how long each answer took. A correct answer that arrives after eight seconds of silence is a counted answer, and counted answers don’t stick.

Watch the clock, not the score

Sit next to your kid for one round and time the answers in your head. Some come back immediately, the way a name does. Others come back after a pause, and their hands or lips will tell you why.

Baker and Cuevas, interviewing twenty four students on single digit multiplication, coded an answer as automatic when it arrived in three seconds or less, and anything slower as a strategy: skip counting, finger counting, the nine times trick on the hands. Twenty four kids isn’t much to build on. But three seconds matches what any parent can see from across the table.

That’s the difference between fluency and memorizing under pressure. A kid who has the fact produces it without effort and can use it inside a long division step without losing their place. A kid rebuilding it every time spends all their working memory on 7 × 8 and has nothing left for the problem it was one step of. Both score the same on a worksheet. Only one of them is done.

So, the rule: a fact your kid gets right slowly isn’t learned yet and should show up again tomorrow. A fact they get right instantly on three separate days can leave the rotation. That’s most of what a fluency program is.

Same time every day, before anybody has to decide

The negotiation is the enemy, not the math. If practice happens whenever there’s time, there’s never time, and every session opens with a conversation you lose about a third of the time.

Pin it to something that already happens at a fixed hour. After breakfast, before the tablet, in the car outside soccer. Mine do it after dinner while the table is being cleared, which isn’t ideal for concentration and is completely reliable, and reliable beats ideal.

Then keep it short enough that it’s not worth arguing about. The day a nine year old works out that they can spend forty minutes complaining about a twenty five minute task is the day the routine dies.

Why short and daily beats long and weekly

Spreading the same total practice across more days beats piling it into one sitting. A 2025 meta analysis in Educational Psychology Review looked at math specifically and found spaced practice beat massed practice with a small to medium effect, g = 0.28 across 27 studies. Worth having for free, since the total minutes are the same either way.

The same review is a useful check on overclaiming. When those authors looked at retrieval practice in math, meaning testing yourself rather than restudying, the average effect was smaller and the confidence interval crossed zero, so the math specific evidence doesn’t settle it. Retrieval practice is one of the best supported ideas in learning science generally, and in math it has fewer studies behind it than the people citing it tend to let on, including people selling math apps.

Spacing is supported. Ten minutes as the correct number isn’t a finding; it’s my number, from watching two children lose interest at slightly different points and rounding down.

One operation at a time, until it’s boring

Don’t mix. Pick addition, or multiplication, or whatever the current wall is, and stay there until those facts come back fast. Mixing early means every question starts with the kid working out what kind of question it is, which costs attention that should go to the fact.

There’s a good reason to mix later: once the facts are fluent in isolation, a mixed round tests whether they really are, because the kid can’t coast on the pattern of the page. That is the exam, not the lesson.

Stop while it’s still going well

This is the part most routines get wrong, including mine for the first year. A worksheet ends when the last problem is answered, which is often several minutes after the kid stopped being able to do it. The day ends in failure, and that becomes what they remember about math practice.

End on a fact they got right. When you can see the answers slowing or the guessing start, stop there and say it out loud: that was a good one to finish on. Which question was last matters more than how many there were. A short fixed count helps, because it usually gets you there before the collapse does, but it’s a ceiling, not a contract.

Never teach during the ten minutes

If they don’t know 7 × 8, tell them 56 and go to the next question.

I know how that reads. It feels like the moment to explain that 7 × 8 is 7 × 4 doubled, and that explanation is worth having. It can’t happen here. The instant you start teaching, ten minutes of drill becomes a twenty minute lesson, and the kid learns that not knowing something extends the session.

Drill time is retrieval time. Teaching time is a different hour and probably a piece of paper. Keep the two apart and both of them get easier.

What to do on a bad day

There will be days when the first three questions go badly and you can see it sliding. Three moves, none of which is pushing through.

Cut it short. Three questions and a clean stop beats ten questions and a meltdown. The routine surviving to tomorrow is the only thing that matters over a year.

Drop back a level. Run facts they already own. That looks like a wasted session and it’s not: easy retrieval is still retrieval, and it ends with your kid feeling competent.

Skip the day, and say so plainly. Not "we will do it later", which becomes a negotiation at bedtime. "Not today, back tomorrow." An announced day off is more durable than four days nobody named.

The failure mode to avoid is a bad session turning into a long session because you kept going to end on a good note. It rarely ends on a good note. It ends at bedtime.

What the app does with all of this

I built PiggyBankMath around my own two kids and it runs on the rules above. A round is ten questions for a seven year old, fifteen up to ten and twenty from eleven, and you can override that. It applies the mastery rule for you: three fast correct answers across two different days, and slow doesn’t count. Mastered facts leave the rotation and come back later. One operation at a time, with a mixed All Math round for the harder test.

The money is the other half. You set a rate per session and a daily cap for each kid, and hand over real cash when you decide to. The app keeps the record and never touches the money.

You don’t need it. Index cards and a phone timer run the same routine, and for some families that’s the better answer because it keeps a screen out of it.

What I’m less sure about

The three second line was measured with a stopwatch, out loud, one question at a time, on a small group of students. A kid typing on a phone is doing something slightly different, and the threshold moves with the input method. I set the app's from the same territory and am not sure it’s right for a seven year old's thumbs.

And all of this assumes the problem is fluency. If your kid is fine on the facts and falls apart on word problems, or is refusing in a way that looks like something other than difficulty, this routine is aimed at the wrong thing.