By Haru · Updated · 2026-09-07
How to Improve Memory — Encoding Beats Effort
Almost everything sold as memory improvement is aimed at the wrong stage. People try to hold on harder at the moment of forgetting, when the decision that mattered was made earlier, when the material went in.
Forgetting is rarely a storage failure. It is usually that the thing was never encoded in a form you could find again — no hook, no structure, no second look. Change what happens at the front and the back takes care of itself, which is why the techniques on this page are all about the first few seconds and the first few days.
A warning about the genre. Training a memory task makes you better at that task, and the transfer to unrelated things is small at best — we say the same in our piece on whether brain training works, and nothing here contradicts it. What follows is not a way to raise a general capacity. It is a set of methods for getting specific material to stick, and those do work.
Four of them carry most of the weight: grouping, placing, retrieving, and spacing.
Group it — the limit is items, not information
Short-term memory holds a small number of items, and the size of each item is up to you. The idea and the word for it come from Miller (1956), whose "magical number seven, plus or minus two" was about items, not amount of information.
A phone number stored as ten separate digits is ten items and beyond most people. The same number as three chunks is three items and easy. Nothing about capacity changed; the packaging did. This is why area codes, card numbers and product keys are printed in groups — the format is doing part of the remembering.
Grouping works best when the chunks mean something. Splitting a number at a year you recognize, a repeated pair, or an ascending run gives each chunk a handle. Arbitrary splits still help, but far less, because you then have to remember the split as well.
Use it on anything with structure. A shopping list becomes three categories rather than eleven items. A password becomes four words rather than sixteen characters. When you try our digit-span exercise, watch whether you are reading digits one at a time or in pairs — the score gap between those two strategies is usually larger than any gap between people.
Place it — put items somewhere you already know
The oldest technique in the field works by borrowing a structure you already have: a familiar route.
Pick a walk you know well — your home from the front door, a commute, a room you can picture. Assign each thing you want to remember to a specific stop along it, in order, as vividly and strangely as you can manage. To recall, walk the route. The order comes free because the route has an order, and the vividness gives each stop a hook.
It feels absurd the first time, and it works anyway, which is the usual reaction. Competitors in memory events use versions of it because nothing else handles long ordered lists as well.
Two practical notes. Reuse the same route for different lists only after the old one has faded, or they interfere. And spend the effort on making each image specific rather than on making it clever — a plain image in a precise location beats a witty one you have to reconstruct.
Retrieve it — testing is not measuring, it is the practice
If you only change one habit, change this one: stop re-reading and start recalling. Roediger and Karpicke (2006) put a limit on when it pays: tested five minutes later, rereading wins; tested days later, the group that had been quizzed remembered substantially more.
Re-reading feels productive because the material gets easier to read each time, and that fluency is mistaken for knowing. Closing the page and trying to produce the material is harder, feels worse, and leaves far more behind. The effort of pulling something out is what strengthens the path back to it.
Do it badly on purpose. Recall before you feel ready, accept the blanks, then check. The gap you notice is doing the work — a failed attempt followed by the answer is one of the strongest ways material sticks, and it beats a comfortable review every time.
This is also why a memory exercise is not only a score. Running a digit span or a sequence task is retrieval practice, and the number it reports is a side effect. Judge a session by whether you were pulling things out, not by whether the number went up.
Space it — the same minutes, further apart
Total study time matters much less than how that time is distributed, and the distribution costs nothing to change. The meta-analysis by Cepeda et al. (2006) pooled 839 comparisons from 317 experiments and found the useful gap grows with how long you need to keep it — days for next week, weeks for next term.
The same hour spread across four days beats an hour in one sitting for anything you need next week. Each return happens after some forgetting has set in, and rebuilding something partly faded is what makes it durable. Study it four times in one evening and the later repetitions land on material that has not faded at all, so they add little.
A schedule that works without any system: revisit the same day, again the next day, again about a week later, again about a month later. Push the interval out each time you succeed and pull it back in when you fail. That is the whole method, and reversing it — cramming as the date approaches — is what most people do.
Combine it with the previous section. Spaced retrieval, not spaced re-reading. The interval decides when, and retrieval decides what you do when you get there.
The ordinary things that beat every technique
Two conditions change memory more than any method, and both are unglamorous.
Sleep is the first, and it is not optional. Consolidation happens during sleep, so material learned before a short night is measurably worse the next day than the same material learned before a full one. A study session that costs you two hours of sleep can be a net loss, which makes late cramming a bad trade twice over.
Attention is the second. Nothing is remembered that was never encoded, and divided attention barely encodes. Studying with a conversation, a feed or a video running produces the feeling of having been there with almost none of the material. This is the mechanism behind walking into a room and forgetting why: your attention was elsewhere at the moment the reason would have been stored.
If memory has got noticeably worse over months, look at sleep, stress and attention before looking for a technique — and if the change is sudden or interferes with daily life, that belongs with a doctor rather than a web page.
What to practice, and how to read the numbers
Practice is more useful when it targets one thing and is measured the same way each time.
Use a digit-span exercise to work on grouping. The strategy is the target, not the score: deliberately read in pairs or threes for a whole session and see where it breaks down. A sequence-following exercise works on order rather than quantity, which is what the route technique is for.
Keep the conditions equal or the numbers say nothing. Same time of day, same length, no second screen. Fatigue moves these scores more than a week of practice does, so a tired session compared against a fresh one tells you about the day rather than about your memory.
And expect the score to move less than the strategy does. The useful question after a session is not whether the number rose but whether you noticed yourself chunking, or placing, or pulling something out before checking. Those are the parts that carry over to the things you actually need to remember.
Sources Miller (1956), "The Magical Number Seven, Plus or Minus Two" (full text) Roediger & Karpicke (2006), “Test-enhanced learning: taking memory tests improves long-term retention”, Psychological Science 17, 249–255 — Europe PMC record Cepeda et al. (2006), “Distributed practice in verbal recall tasks: A review and quantitative synthesis”, Psychological Bulletin 132, 354–380 — Europe PMC record
Frequently asked questions
Which exercise should I start with?
The digit-span one, because it makes chunking visible. Run it once reading digits one at a time and once deliberately grouping them into pairs; the difference between those two runs is usually larger than any difference between people, and seeing it is what makes the technique stick. Add a sequence exercise afterwards when you want to work on order.
Is it free, and do I need an account?
The exercises are free, need no account, and nothing is uploaded — the counts stay in your browser and are discarded when the tab closes. That also means no history is kept for you, so write your own number down after each session if you want to see a trend.
Does memory training make you smarter in general?
No, and this page does not claim it. Practicing a memory task reliably improves that task and transfers weakly at best to unrelated abilities. The techniques here are worth learning because they get specific material to stick, which is a narrower and more achievable promise than raising a general capacity.
I forget names immediately. What works for that?
Names are hard because they are arbitrary — nothing about a face implies a particular word. Repeat it out loud once in the conversation, connect it to something concrete you can picture, and retrieve it deliberately a few minutes later rather than hoping it stayed. That is grouping, placing and retrieval applied to the smallest possible item.
How long before any of this shows?
Chunking helps within a single session because it changes what you are doing right now. Spacing and retrieval show over weeks, since both are about what survives between sessions rather than what happens inside one. If you need a result tomorrow, the honest answer is to sleep properly and use retrieval rather than re-reading.
My memory has got worse recently. Is that normal?
Short-term changes usually track sleep, stress, attention and how much you are trying to hold at once, and those are worth examining first. But a web page cannot tell you which case you are in, and a change that is sudden, one-sided, or interfering with everyday life is a reason to see a doctor rather than to practice harder.