Learn to diagnose the mistakes in a memory training run, from encoding slips to swapped positions, and turn each one into a specific fix.
Most people finish a memory drill, glance at the score, and start the next one. That is a fast way to practice and a slow way to improve. Your memory drill results contain far more information than a single number: they show where your system holds, where it leaks, and which kind of mistake is costing you the most. This guide teaches you how to read a result, sort your errors into a handful of types, and match each type to a fix.
A score tells you how you did. It does not tell you why. Two people can both score 120 in a five-minute numbers drill and need completely different advice: one is memorizing too slowly but recalling perfectly, and the other is memorizing plenty and losing rows to careless slips. Treating those two people the same would waste both of their time.
Scoring rules make this even more important. In the number events, a row with one mistake earns half points and a row with two mistakes earns nothing, so the distribution of your errors matters more than their total. Ten mistakes spread across ten rows costs you half of ten rows. Ten mistakes packed into five rows costs you all five. You can only see that difference if you look at the result row by row. If the scoring rules are new to you, start with the memory championship scoring rules.
Before you classify anything, look at the position of each mistake. Patterns in position are the fastest clues you will get.
Write the pattern down in one line. It becomes the first line of your practice notes.
Once you know where the errors are, look at what each one actually is. Almost every error in a memory drill falls into one of the following categories.
You turned the information into the wrong image in the first place. A pair of digits became the wrong person, or a word became a picture that does not really match it. When you recall, you retrieve the image faithfully, and it decodes into the wrong answer.
How to spot it: the wrong answer is a plausible neighbour of the right one, such as a digit pair that sounds or looks similar under your system.
Fix: drill your image conversions on their own, away from any palace, until they are automatic. If you use a number system such as the Major System or PAO, encoding slips almost always mean the system is not yet fast enough for the speed you are attempting.
You made the right image but stored it in the wrong place, or you stored two images in the same location. In a card drill this appears as two cards swapped, or a card that appears one position too early.
How to spot it: the answers you wrote are correct but in the wrong order, and one wrong item often has its partner elsewhere in the row.
Fix: slow down your memory palace walk so that every image gets a clear moment in its own location. Check that your locations are distinct enough and that you always travel the route in the same direction.
A specific kind of placement slip in which two neighbouring items exchange places. Under the card-event rules it is worth remembering that swapping two cards counts as two errors, not one, so a single moment of confusion can be expensive.
Fix: add a small directional cue to every image, such as always placing the first item of a pair on the left of a location and the second on the right.
You know you memorized something and cannot pull it back. The location is there, the image is not. In a scored event a blank is usually as costly as a wrong answer, and in numbers it counts as a mistake.
How to spot it: you remember the feeling of the image but not the picture, or you leave gaps you can describe afterwards.
Fix: make images larger, stranger and more active. A faint image is the usual cause. Also practise recalling in a calm state: rushed recall makes blanks more likely.
You drifted. A whole stretch of the sheet goes in poorly, with no obvious system fault. Tiredness, a noise, or a stray thought is typically the cause.
How to spot it: a run of consecutive errors in one region, with the rest of the sheet clean.
Fix: shorten practice sessions and train under mild distraction. It is also worth noting how you felt before the drill, which we come back to below.
You memorized the right material but misjudged where one chunk ends and the next begins. A three-digit group is treated as two digits, and everything after it shifts by one. The result is a whole row that is nearly right but offset.
Fix: decide your chunk size before you start and keep it fixed. Our guide to chunking covers how to choose one.
You remembered the right thing and wrote it wrong. This is specific to words and names. In Random Words, a clearly memorized but misspelled word loses only that word rather than halving the column, so it is the cheapest kind of error, but it is entirely avoidable with a spelling check in your routine.
One drill is noise. Three drills of the same event begin to show a signal. After each run, tally your errors by type. A tiny table on paper is enough: one column per error type, one row per run.
After three or four runs, one column will usually dominate. That is what to fix, and nothing else. It is tempting to work on everything at once, but changing your image conversion, your palace pacing and your recall routine in the same week makes it impossible to tell which change helped.
The score cannot record whether you slept badly, whether you tried a new image for one pair, or whether you switched to a bigger chunk. A short private note attached to the run can. Good notes are brief and factual: "lost the thread at row 12", "new image for 42 held up", "tired, probably should not count". MemoChamp lets you add a private note to a saved run for exactly this reason, and reading your notes back after a month is often more instructive than the scores themselves.
Pick the dominant error type and choose one adjustment. Then run the same event again under the same conditions, and see whether that type of error shrinks. If it does, keep the change. If it does not, drop it and try the next idea. Over a few weeks this loop builds a personal playbook that no general guide can replace.
Imagine a five-minute Speed Numbers run in which you score 150 points, and the sheet shows six rows attempted. Two rows are perfect, two have one mistake each, one has two mistakes, and the last row is partial. By the scoring rules, the perfect rows are 40 points each, so 80, the single-mistake rows are 20 each, the two-mistake row is zero and the partial row, with 30 digits correct, adds 30.
Looking at the mistakes, you find that the two-mistake row has both errors in the same half of the row, and that one of the single-mistake rows is also in that stretch. That suggests a damaged section of your route rather than a random accuracy problem. The right response is to walk that part of the palace on its own, check that each location is distinct, and replace any that are not. The wrong response would be to slow the whole run down, which would cost you speed everywhere to fix a problem in one place.
Reviewing takes five minutes and improves everything that follows. Make it the last step of every session: score, pattern, error type, note. If you are training for a specific event, the drills in training show each row of a result, so the first two steps take almost no effort. And if you are choosing which event to work on, the guide to the ten disciplines is a good place to start.
