Ask a child to add 3/4 and 1/2 and you may get 4/6. Most adults see that and think careless. It isn't. That answer is the product of a rule the child believes: add the tops, add the bottoms. It is consistent, it is applied confidently, and it will produce a wrong answer every single time until someone finds the rule and replaces it.
That is the thing fourteen years of classrooms taught me more than anything else. Children very rarely make random mistakes. They make systematic ones, driven by a mental model that made sense somewhere earlier and then got over-applied. Researchers call these misconceptions. Parents usually call them "he keeps doing that thing again."
Why a score can't find them. A 73% tells you how many questions went wrong. It cannot tell you whether a child missed six subtraction problems because she doesn't know her facts, because she regroups from the wrong column, or because she subtracts the smaller digit from the larger one regardless of which is on top. Those are three completely different lessons. Re-teaching subtraction in general will fix none of them efficiently, and the child will conclude she is simply bad at math.
Misconceptions you have probably already seen. Bigger denominator means bigger fraction. Multiplication always makes a number larger. The equals sign means "write the answer here" rather than "both sides balance." Longer number means greater value, so 0.45 beats 0.7. In reading, a child decodes every word correctly and still cannot tell you what happened, because he believes reading means saying the words, not building a picture. In science, heavier things fall faster. None of these are silly — each one is true in the narrow set of examples where the child first met it.
How to find one at your kitchen table. You don't need software for this. Ask the child to talk you through the problem out loud: "Show me how you got that." Then give the same skill two more times with the numbers changed. One wrong answer is noise. The same wrong logic three times in a row is a pattern, and the pattern is the thing you teach to. Watch for what stays constant across the errors — that constant is the rule in their head.
Then teach the contrast, not the correction. Telling a child "no, bigger denominator means smaller pieces" rarely sticks, because their rule still works on the examples they remember. What works is a case their rule cannot survive: cut one pizza into 3 pieces and one into 8, and hand them a slice of each. Misconceptions get replaced by evidence, not by volume.
This is exactly what we built into Lesson Quest. When a child misses a tagged question in the Adaptive Lesson Builder, it doesn't just serve an easier problem. It asks up to three more questions built to test that one specific reason, and only when the same reason shows up again does it say anything — because one miss proves nothing. Confirmed patterns then appear on the Learning Patterns page in plain language: not a percentage, but a sentence like "she is adding numerators and denominators together," plus what to do about it. Our free standards-aligned lesson plans are a good place to get the practice problems you need for that.
The bottom line. A wrong answer is data, not a verdict. Once you stop asking "how many did she miss" and start asking "what does she believe that made this answer look right," teaching gets faster, kinder, and far more precise. For more on what that should mean in the tools you use, read what "adaptive" tutoring should actually mean.