Two children answer 47 + 38. One says "85" in two seconds. The other stares at the ceiling for twenty, lips moving slightly, and then says "85."
Same question. Same answer. Which child is better at maths?
It is a trick question — you cannot tell yet. And the reason you cannot tell is the whole point of this article. Speed, on its own, tells you almost nothing about a child's maths. What matters is what was happening during those twenty seconds — and there are versions of that story that should reassure you completely, and one version that deserves your attention.
Three things that get blended into one word
When a school report says "needs to work on speed," or a child comes home saying "everyone finishes before me," three quite different things are being folded together.
Accuracy is the first: do they get it right? Not quickly — at all. Given time and calm, does the method hold up and the answer come out correct?
Fluency is the second: is it automatic? A fluent child doesn't work out 8 + 7; they know it, the way they know their own name. No fingers, no counting, no ceiling-staring. The answer is simply there.
Speed is the third: how fast can they perform under pressure — a ticking clock, a class race, an exam hall? Speed is the only one of the three that anyone can see from across a room, which is why it gets all the attention. It is also the least informative of the three, because it is downstream of the other two.
A child can be accurate but not fluent (right answers, slowly assembled). A child can be fast but not accurate — and this combination, which timed drills quietly manufacture, is the worst of the four. What no child can be is genuinely fast without being fluent first.
The order that matters
Here is the rule the rest of this article hangs on: accuracy comes first, then fluency, then speed — in that order, and never the reverse.
Speed is not a skill you train directly. It is a by-product. When a fact has been retrieved correctly enough times that it becomes automatic, speed arrives on its own, unforced, the way a well-practised piano piece speeds up without the pianist deciding to play faster.
This is not just a comforting reframe; it is roughly how the field defines the goal. The National Council of Teachers of Mathematics describes procedural fluency as a combination of accuracy, efficiency and — the part parents rarely hear — flexibility: choosing sensibly among methods, not just executing one method fast.
Notice what is missing from that definition: raw speed under pressure. A stopwatch measures none of the three components. It measures the shadow they cast.
Why timed drills on a shaky foundation backfire
If speed is the visible product, the tempting shortcut is to train the visible product: print the timed sheets, start the stopwatch, race the sibling. And if the underlying facts are already secure, a little friendly time pressure does no harm.
But if the facts are not secure — if 8 + 7 is still being counted rather than known — the stopwatch does something destructive. Pressure plus insecure facts produces guessing. The child learns that when the clock is running, an answer, any answer, beats no answer. "Fast but wrong" becomes a habit, and it is a far harder habit to unlearn than slowness, because it looks like progress on the page while quietly corroding the child's relationship with being correct.
Speed is what fluency looks like from the outside. Chasing the outside first gets you children who are fast at being wrong.
There is a second cost. The researcher Jo Boaler has argued for years that premature timed testing is one of the most reliable triggers of maths anxiety in young children — the racing heart, the blank mind, the "I'm just bad at maths" verdict delivered by a seven-year-old about themselves.
Whether or not every child is affected equally, the ordering claim underneath is uncontroversial: pressure applied before fluency exists doesn't create fluency. It creates performance anxiety about the absence of it. If your child already tenses at the sight of a maths book, our piece on maths anxiety in children walks through that spiral — and how to unwind it.
When slow is fine — and when it's a flag
So when should twenty seconds of ceiling-staring worry you? Mostly, it shouldn't. Slowness is fine — often a good sign — in three situations.
The topic is new. A child meeting equivalent fractions for the first time should be slow. Slow here means thinking.
The problem is genuinely hard. A multi-step word problem rewards the child who reads it twice. Some of the best mathematicians in any classroom are the deliberate ones at the back, still checking.
The child is a careful checker. Some children re-verify before committing. That is a temperament, not a deficit — and in an exam hall full of rushers it is frequently worth marks, not costing them. (Rushing has its own failure mode; we've written about why so-called careless mistakes are rarely careless.)
The one pattern that is a flag: slowness on small, basic facts the child has met hundreds of times. If 6 + 7, 13 − 8 or 4 × 5 still takes visible effort in Grade 4 (Class 4 in India, Year 5 in England), that is not a thinking child. That is a fluency gap — a fact that never made the journey from "worked out" to "known," so it gets rebuilt from scratch, every single time, forever.
One mother we heard from, near Melbourne, had a son whose school had settled on the label "slow." She sat with him one evening and asked him to talk her through 6 × 7 out loud. What came back was remarkable: "six fives are thirty, six twos are twelve, thirty and twelve is forty-two." Correct — and completely reconstructed from parts, every time, for a fact his classmates simply knew. He was doing this for nearly the whole multiplication table: one hundred per cent accurate, and deriving everything from scratch like a mathematician marooned without a memory. Clever, and exhausting. Eight weeks of short daily fact-retrieval practice later, the "slow" label was gone. Nothing about his understanding changed. It was always there — it just finally got to travel light.
That story is worth sitting with, because it shows what the label "slow" so often hides: not a weaker child, but a child paying full price for facts other children get for free. And like most maths struggles, the real cause sat one layer below where anyone was looking — the same pattern we describe in why children struggle with maths, where the visible problem is almost never the actual problem.
How to build genuine speed
The good news: the slow-but-accurate child is in the best possible starting position. Accuracy — the hard part, the understanding — is done. What remains is consolidation, and consolidation responds beautifully to a specific kind of practice.
Retrieval, not review. Roediger and Karpicke's 2006 experiments showed that testing yourself on material beats re-reading or re-studying it for long-term retention — often dramatically so.
For maths facts, that means the child answering — flashcards, oral quizzing, a short mixed quiz — rather than staring at a filled-in tables chart. The slight effort of dragging the answer out of memory is not a side effect. It is the mechanism.
Tiny daily doses. Five to ten minutes a day beats an hour on Sunday, for reasons of memory rather than discipline — we go into the numbers in how many minutes of daily practice is enough. Little, often, and low-stakes.
Strategies before drill. Before drilling 6 × 7, make sure the child can derive it (double 3 × 7; or 6 × 5 plus 6 more). A derived fact recalled a few dozen times becomes a known fact. A drilled fact with no structure underneath becomes a guess under pressure.
Time pressure last, and gently. Only once facts are coming back reliably and unaided — no fingers, no reconstruction — is it time for soft time pressure. "Beat your own time from Tuesday" builds fluency into speed. "Beat the rest of the class" builds dread.
That ladder — accuracy, then retrieval practice until fluent, then gentle pace — is exactly the ladder a good adaptive system climbs on a child's behalf, adjusting the rung daily.
The question to ask instead
So, back to our two children and 47 + 38. Before deciding anything about the twenty-second child, you would want to know: was that twenty seconds of strategy — new-ish problem, sensible method, careful check — or twenty seconds of rebuilding 7 + 8 from the ground up? The first needs nothing from you but patience. The second needs a few weeks of the right practice, aimed at the right facts.
If you would rather not diagnose that by hand, this is precisely what CREST Champs' free daily adaptive practice does quietly in the background: it notices which facts are known and which are being reconstructed, and it feeds the child ten minutes a day of exactly the retrieval work those facts need — accuracy first, fluency next, and speed arriving on its own, the way it was always meant to.