How to remember what you watch on YouTube

Why do you forget videos so quickly?

Because watching a video is a recognition task, and remembering is a retrieval task. They are not the same skill, and practising one does very little for the other.

While the video plays, every idea arrives with its full context attached — the speaker's framing, the example, the diagram on screen. Each sentence feels obvious as it lands. That feeling is real, but it is a measure of how easy the material is to recognise while it is in front of you, not of whether you could produce it later from an empty page.

Hermann Ebbinghaus documented the shape of what happens next in 1885, and it has held up: retention drops steeply soon after learning, then the curve flattens. The steep part is the problem. A three-hour podcast you found genuinely brilliant on Sunday is, by the following weekend, a vague sense that it was about incentives and maybe something about a bridge.

Does taking notes while watching help?

Less than you would expect, and for two reasons.

The first is attention. Typing a sentence while listening to the next one splits your attention across two tasks, and the notes usually win — you end up with a decent transcript and a poor memory of what any of it meant.

The second is what the notes are for. Notes taken during a video are a capture artefact. They record that an idea passed by. Nothing about writing something down obliges you to ever retrieve it again, and an unretrieved note decays on the same curve as the video did. This is why a read-it-later app fills up and stops being opened: it is a capture system with no recall step, so it accumulates rather than compounds.

Notes are worth taking. They are just the beginning of the process, not the whole of it.

What actually works?

Three things, all of which feel worse in the moment than rewatching does — which is exactly why most people skip them.

1. Retrieve instead of rewatching

Close the tab and try to state what the video argued. Out loud, or in two sentences on paper. The struggle is the mechanism, not a sign it is going badly. Roediger and Karpicke's work on test-enhanced learning found that being tested on material produced substantially better long-term retention than spending the same time restudying it — and, notably, that students consistently predicted the opposite.

That prediction error matters. Rereading and highlighting feel productive because they are fluent. In the large review by Dunlosky and colleagues, both were rated low utility, while practice testing and distributed practice came out on top.

2. Space the attempts out

Reviewing something three times in one evening is worth far less than reviewing it three times across three weeks, even though it costs the same total minutes. Spacing forces partial forgetting between attempts, and reconstructing a half-faded memory is what makes it durable.

3. Connect it to something

An idea with one route to it is fragile. An idea linked to three others you already hold has three routes, and is far more likely to surface when you actually need it. In practice this means asking, for each thing worth keeping: what does this contradict, confirm, or explain?

How long should the gaps be?

Long enough that recall takes effort, short enough that it still succeeds. A schedule roughly like this works for ideas rather than facts:

The exact numbers matter less than the shape: each successful recall earns a longer gap, and a failed one shortens it. That expanding interval is the whole idea behind spaced repetition, and it is why a fixed "review everything every Sunday" habit underperforms — it spends the same effort on things you already know cold and things you are about to lose.

How do you do this without it becoming a second job?

This is where the honest answer is that most people cannot, unaided. The method is not difficult, it is just relentless: something has to decide what is due, on what day, and put it in front of you. Doing that bookkeeping by hand is more work than the reviewing itself.

1. Be ruthless at capture. Most videos do not deserve a review schedule. One or two ideas per week, held properly, beats fifty saved and none retained.
2. Write the retrieval prompt at save time, while you still remember why it mattered. "What did he say the real constraint was?" is a prompt. "Interesting video on supply chains" is not.
3. Let something else own the calendar. The scheduling is the part that fails first, and it is the part a tool can fully take over.
4. Put reviews where you already are. A prompt in a chat app you check anyway gets answered. A dashboard you have to remember to open does not.

That last point is the one people underrate. The failure mode of every personal knowledge system is not that the method was wrong — it is that opening the app became one more thing to remember, and the system quietly died of friction.

How Recl handles this

Recl is built around exactly the gap described above: it assumes capture is easy and recall is the hard part. You save a video, article or PDF; it produces a structured summary with key takeaways, builds quiz questions from the material, and then sends review prompts through Telegram or Discord on an expanding schedule. Marking one reviewed pushes the next one further out.

The design choice worth naming is the delivery: reviews arrive in a chat app you already have open, because a review you never see is a review that did not happen.