Every fact timed separately, so the result is a list of what to practise rather than a number
A score out of thirty tells you very little. This times every fact on its own and reports which ones took longer than three seconds or came out wrong, because those are the facts costing you time inside every longer calculation. Nothing is saved and nothing leaves your browser.
Most people do not have a times-table problem. They have a four-fact problem, hidden inside an average. Getting 27 out of 30 looks fine and tells you nothing about which three to work on.
Three seconds is the line this check uses. A fact you can produce inside a second costs nothing when it appears in the middle of a longer sum. A fact that takes four seconds costs four seconds every single time, and worse, it takes attention away from whatever the multiplication was part of.
That is also why the facts are shuffled. Asked in order, the previous answer cues the next one and everything looks fluent.
Take the list of slow facts and drill only those. A minute a day on four facts will fix them within a fortnight; a minute a day on all 66 will not.
Attach each slow fact to one you already have. If 6 × 7 keeps slipping but 6 × 6 = 36 is instant, the route in is 36 + 6. It is slower than recall at first, and repetition turns it into recall.
Re-run the check a fortnight later. The list should be shorter and the median time lower. If neither has moved, the drilling is happening in order rather than shuffled.
The middle of the grid is where it bites: 6, 7 and 8 multiplied by each other. They have no digit pattern to lean on, so unlike the 9s, the 11s or the 5s they cannot be reconstructed from a rule — they simply have to be known.
Because the grid is symmetrical, this check asks each pair once. Knowing 7 × 8 is knowing 8 × 7, so there are 66 facts to worry about in the range 2 to 12, not 121.
There is no pass mark here and no comparison to anyone else. Reaction times vary with sleep, caffeine, keyboard and how warmed up you are, so a single run is a snapshot rather than a measurement.
What is worth tracking is your own list getting shorter over a few weeks. That is a real change; a single day's score is mostly noise.
Under about a second for a fact that is genuinely automatic. This check flags anything over three seconds, which is the point at which a fact is being worked out rather than recalled.
The 6, 7 and 8 rows multiplied by each other. They have no digit pattern to lean on, so they cannot be reconstructed from a rule the way the 9s and 11s can.
Because multiplication is commutative, so 7 × 8 and 8 × 7 are the same fact. Asking both would double the length of the check without testing anything extra.
No. The check runs entirely in your browser, nothing is stored and nothing is sent to us. Closing the page loses the result.
Every couple of weeks. Day-to-day variation is large enough that a single run tells you about today rather than about your progress.