You cannot retain 90 percent of what you learn by teaching it, and the chart that says you can was never based on research. The "learning pyramid" percentages (5 percent from lectures, 10 percent from reading, 90 percent from teaching others) have no traceable source. What does raise retention is measurable: retrieval practice, spaced review, and explaining ideas in your own words. Those three reliably move recall from roughly 40 percent to roughly 60 percent a week later.
Where the 90 Percent Claim Comes From
Search for how to retain more of what you learn and you will meet the same triangle within about thirty seconds. It is usually called the Learning Pyramid or the Cone of Learning, and it assigns a tidy retention rate to each way of studying:
| Activity | Claimed retention | Evidence |
|---|---|---|
| Lecture | 5% | None |
| Reading | 10% | None |
| Audiovisual | 20% | None |
| Demonstration | 30% | None |
| Discussion group | 50% | None |
| Practice by doing | 75% | None |
| Teaching others | 90% | None |
Every version credits the same institution: the NTL Institute for Applied Behavioral Science in Bethel, Maine. And here the trail ends. When researchers asked NTL for the underlying study, the institute acknowledged it could not produce it. The original data, they explained, was no longer available.
Look at the numbers again and the problem becomes obvious. They are all multiples of five. They sum with suspicious neatness. Real retention data is never this clean, because retention depends on what you learned, how long ago, how it was tested, and who you are. A single number for "reading" is not a finding. It is a slogan.
The Chart Was Built on a Diagram That Had No Numbers
The pyramid is often traced to Edgar Dale, an education researcher who published his Cone of Experience in 1946. Dale's cone was a genuine and useful idea: it arranged learning experiences from direct and concrete at the base to abstract and symbolic at the top.
Dale attached no percentages to it. None. He explicitly warned readers not to treat the cone as a ranking of effectiveness. The numbers were bolted on later by other people, and once attached, they spread on their own.
Kare Letrud and colleagues have tracked this diffusion through the academic literature, documenting how the fabricated percentages were cited, re-cited, and laundered into legitimacy across decades of education papers, textbooks, and corporate training decks. The citations often point at each other in circles. Nowhere do they terminate in an experiment.
So How Much Do You Actually Retain?
There is no single answer, and that is the honest part. Retention depends on the material, the delay, and the test. But we do have solid numbers from controlled experiments, and they look nothing like the pyramid.
In a 2006 study by Henry Roediger and Jeffrey Karpicke, students studied a prose passage and were split into groups. One group reread the passage four times. Another read it once, then took three recall tests with no feedback.
Tested five minutes later, the rereading group did slightly better. Tested a week later, the picture reversed completely. The repeated-testing group recalled about 61 percent of the material. The repeated-reading group recalled about 40 percent.
That is a 21-point gap from changing nothing except whether students pulled the information out of memory or pushed it back in. Both groups spent the same time. One method simply worked better.
The detail worth sitting with: students in the rereading group predicted they would do better. Fluency feels like knowledge. It is not.
Does Teaching Others Actually Help?
Yes, and this is why the myth survives. The pyramid's top row points at something real, then attaches a fabricated number to it.
Teaching does improve retention. Work by Nestojko and colleagues found that simply expecting to teach material, before any teaching happens, improved how well students organized and recalled it. The anticipated obligation changed how they read. They looked for structure, main points, and the connections they would need to make explicit.
The mechanism is not magic. Explaining forces retrieval, exposes gaps, and demands that you organize scattered facts into a sequence someone else can follow. Those are three separate memory benefits stacked into one activity.
What the research does not support is 90 percent. Nobody measured that. Teaching helps meaningfully; it does not make you nearly perfect.
What Actually Raises Retention
Three techniques have survived decades of replication. They are unglamorous and they work.
- Retrieval practice. Close the source and recall the idea before checking. This is the single most reliable memory intervention in the literature, and it is the one most people skip because it feels harder than rereading.
- Spaced review. Revisit after a day, a week, and a month. A 2006 meta-analysis by Cepeda and colleagues synthesized 254 studies covering more than 14,000 participants and found spaced practice beat massed practice at identical total study time.
- Elaboration. Explain the idea in your own words and connect it to something you already know. This is where the teaching effect genuinely lives.
Notice what is absent from that list: highlighting, rereading, and watching. Not because they are worthless, but because they are input methods, and retention is built by output.
The Honest Version of the Pyramid
If you want a mental model that keeps the useful intuition without the invented numbers, use this one:
| What you do | Mental mode | Effect on retention |
|---|---|---|
| Read, watch, listen | Passive | Necessary input. Weak on its own. |
| Highlight, take verbatim notes | Active | Slightly better. Still mostly copying. |
| Summarize in your own words, self-quiz | Constructive | Strong. You are generating, not receiving. |
| Explain to someone, argue, get questioned | Interactive | Strongest. Gaps surface immediately. |
This ordering comes from Michelene Chi's ICAP framework, which is grounded in actual experiments rather than a lost memo. It preserves the pyramid's one good instinct, that doing beats receiving, and drops the fake precision.
A Realistic Retention Target
Stop aiming at 90 percent. Aim at this instead: learn less, recall it more often.
One concept a day, retrieved from memory the next day, reviewed a week later, and explained out loud once, will outperform an hour of highlighting by a wide margin. Not because of a percentage on a chart, but because every step in that sequence is an output step.
If you want that loop to run without you designing it, that is exactly what a daily general knowledge system is for: short lessons, a quiz at the end of each one, and topics that rotate so breadth builds on its own. The quiz is not decoration. It is the part that does the work.
For the full set of guides and quizzes, start from our general knowledge hub.
Bottom Line
The 90 percent figure is folklore. It has been repeated for so long, by so many credible-looking sources, that it now feels like something everyone knows. It is not. Nobody ever measured it.
What is measured, repeatedly and across decades, is smaller and more useful: test yourself instead of rereading, space your review, and explain things out loud. Those three will not get you to 90 percent. They will roughly halve what you forget, which is a result you can actually bank.
We cover retrieval, spacing, and the daily loop in this six-minute video as well.
Frequently Asked Questions
How to retain 90% of what you learn?
You cannot, and the claim is not based on research. The 90 percent figure comes from the learning pyramid, a chart credited to the NTL Institute that has no traceable source study. Real retention gains come from retrieval practice, spaced review, and explaining ideas in your own words.
Is the learning pyramid real?
The diagram is real but the numbers are not. It is loosely based on Edgar Dale's 1946 Cone of Experience, which contained no percentages and was never meant as a ranking of effectiveness. The retention figures were added later by unknown authors and have never been traced to an experiment.
Do you really remember 90% of what you teach others?
Teaching does improve retention, but not to 90 percent. Research by Nestojko and colleagues found that even expecting to teach material improved recall, because it changes how you read. The effect is real and worth using. The specific percentage was invented.
What percentage of what we read do we actually remember?
There is no single number, because retention depends on the material, the delay, and the test. In Roediger and Karpicke's 2006 experiment, students who reread a passage four times recalled about 40 percent of it a week later, while students who self-tested recalled about 61 percent.
What is the most effective way to remember what you learn?
Retrieval practice: close the source and recall the idea before checking. It is the most reliably replicated memory technique in the research literature, and it consistently outperforms rereading and highlighting at identical study time. Pair it with spaced review for the largest effect.
📚 Keep Learning
Build the Retrieval Habit, Not the Highlight Habit
Every NerdSip lesson ends with a quiz, because recalling an idea is what makes it stick. Five minutes a day, across any topic you are curious about.
