What you teach all at once, they forget all at once

Take a fixed amount of studying — say an hour — and you can spend it two ways. Pour it all into one sitting, or break it into shorter sessions spread across several days. Same material, same total minutes. For well over a century we’ve known the spread version wins, and it isn’t close: what you learn in spaced sessions is remembered dramatically longer than the identical content crammed into one block. This is the spacing effect, and it’s about as close to a law as soft science gets — it holds across ages, subjects, motor skills and facts, and it survives nearly every attempt to make it go away. The strange part is how strongly it cuts against instinct. Cramming feels productive. Massed practice gives you a lovely sense of fluency by the end of the session — and that fluency is mostly an illusion that evaporates by the time you need the thing.

Why would memory work this way? The leading account is that each time you retrieve or re-encounter something after a delay — long enough that it’s begun to fade — your brain does more work to bring it back, and that effortful re-access is precisely what tells the system “this one matters, file it deeper.” Retrieve it again while it’s still fresh and there’s no work to do, so no signal is sent. Forgetting a little, then recovering, is not a bug in the process; it’s the mechanism. Which means a small amount of forgetting between exposures isn’t the enemy of learning — it’s the ingredient. Ebbinghaus’s famous forgetting curve, the steep drop-off of memory over time, is only half the picture; the other half is that each well-timed review makes the next drop-off shallower.

Now look at how most products onboard, and wince. The default is a first-day firehose: a guided tour, a checklist, tooltips on every button, a “here’s everything our product does” the moment someone signs up — the entire education compressed into the one session where the user has the least context to hang it on. It’s a cram. It feels thorough, it demos well, and it’s forgotten by the second visit, which is exactly why so many products watch users return on day three and behave like they’ve never seen the place. The build decision the spacing effect argues for is almost the opposite of the firehose: teach a little, let them leave, and teach the next thing on the next visit. Get them to one real win on day one and stop. Reintroduce the second capability when they come back and actually have a reason for it. Time your re-engagement — the email, the nudge, the “here’s a thing you haven’t tried” — to land in the dip of the forgetting curve, when a small reminder does the most work, and let the intervals stretch as the memory hardens. That’s the same principle the best flashcard apps run on; it works just as well on a product’s features as on vocabulary.

Two honest cautions. First, spacing trades a feeling for a result: spaced learners feel less fluent in the moment than crammers do, so users (and the teammates reviewing your onboarding) may report that the drip feels slower or less complete — you’re buying week-four retention with a little day-one comfort, and you have to actually believe that trade to ship it. Second, “spread it out” is not a license to nag. The mechanism rewards a reminder timed to a genuine need or a real forgetting dip, not a stream of notifications; space the teaching, not the pestering, and make each spaced touch carry something the user is glad to be reminded of.

The science, to look up: the spacing effect, first observed by Hermann Ebbinghaus (1885) alongside his forgetting curve; the large meta-analysis of distributed practice by Cepeda, Pashler and colleagues (2006); and spaced-repetition scheduling (the Leitner system and its modern descendants). This is one of the most replicated findings in learning science — safe to state plainly — though the optimal gap depends on how long you need the memory to last, so treat “expanding intervals” as a rule of thumb, not a precise formula.