Every learning app for young children shows you a number, and almost all of those numbers are decoration. Seventy-eight per cent. Level four. A bar that is two-thirds full. They feel like information and they are mostly a mood.
We decided early that our progress screen would not show a percentage anywhere, and it has been one of the more useful constraints in the whole product. Here is the reasoning, and what we show instead.
Why a percentage on a four-year-old is a made-up number
Say your child answers eight of ten questions correctly. Seventy-eight per cent, eighty per cent, whatever the rounding — what does it tell you?
Not much, because you do not know:
- Whether the questions were the same difficulty. Two of them may have been the same question in different clothes.
- When they were answered. Six on Monday when they were fresh, four on Thursday at half past six.
- What “wrong” meant. A tap on the wrong picture, a mumble the microphone could not make out, or a child who put the phone down and wandered off.
- How many there were. Eighty per cent of five is a coin toss with extra steps.
Then the number gets averaged with other numbers, and by the time it reaches a dashboard it is a number about numbers. A parent reads “78%” as “my child knows about three-quarters of this”, which is not a claim anybody can defend.
The deeper problem is direction. A percentage is designed to go up. Once a screen exists whose job is to go up, everything else bends towards making it go up — easier questions, more forgiving scoring, half marks. We would rather not have the pressure in the building.
The rule we use instead
A skill counts as consistent only when your child has got it right in at least nine of the last ten times it came up — and those ten have to be real, graded attempts, on the current difficulty band, spread across days rather than crammed into one session.
Nine of ten is deliberately strict. Four-year-olds guess, and a child who guesses their way to seven of ten has told you nothing. Nine of ten across several days is a claim we are willing to make out loud.
The nice side effect: it can go down. If a skill was consistent in March and is not in May, the screen says so. A number that can only rise is not measuring anything.
“Practised” is not “wrong”
Speaking games are where scoring goes wrong most often. A three-year-old says a word, the microphone gets an ambiguous sound, and the app has to decide.
Marking that wrong would be dishonest — most of the time the child said it perfectly and the room was noisy, or they are three and that is what a three-year-old’s “elephant” sounds like. So unclear speech scores as practised: it counts as something they did, it appears in the totals, and it never counts against them. The same applies when a child quits halfway through a game, or a timed moment runs out. They practised. That is a true statement.
On the screen this shows up as a line like “Practising · 4 tries this week” where a progress bar would be a lie. A row that admits it does not know enough yet is more useful than a bar at 60%.
What we do show
- Per skill, not overall. One bar per skill, and only for skills that actually have data. No invented ranking across the whole child.
- The status in words. Consistent, or still practising, with the number of tries behind it.
- The last ten, honestly. The rule above, applied and named.
- What happened, in order. A plain feed: this lesson, this song, this story, on this day.
- Stars. The child’s own count of things they finished. Not a currency, never spent, never lost, and it does not measure whether they were right.
- A weekly recap in plain words — a short written summary of what the week actually held.
And on the child’s side of the app there is no progress screen at all. Mastery is a grown-up concept. A four-year-old does not need a dashboard telling them how they are doing; they need to be told they finished, and then to be allowed to stop.
How to read it as a parent
Three habits make this kind of screen useful rather than anxiety-inducing:
- Look weekly, not daily. Daily is noise. A four-year-old has bad Tuesdays.
- Read the practising rows first. They are where the useful five minutes is. If a skill has plenty of tries and no consistency yet, that is the one to play a game about.
- Ignore totals. “Fourteen activities this week” is a fact, not an achievement, and comparing it with last week’s number will only make you cross with a small child.
If you want a way to act on what you see, the five-minute daily loop in teaching letter sounds at home is the shape that fits around it.
The uncomfortable bit
Honest measurement means the screen sometimes has nothing to say. A week where your child played two songs and no lessons produces almost no learning data, and we show that rather than manufacturing something.
That is the trade. You get a number you can trust, and you get fewer of them.
Make this in Dinomy
Progress in Dinomy is built from what your child actually did — every answer in every activity is one recorded event, tagged with the skill it belongs to, and every bar on the screen is derived from those events rather than from anything the AI wrote. Per skill, no percentages, “practised” never counted as wrong, consistency only at nine of the last ten, and a weekly recap in plain words you can read in a minute. Dino, the helper on the parent side, will explain any of it and suggest what to make next — it advises, it never creates anything or spends anything on your behalf.
