Make It Stick Summary and What Still Holds Up

What does Make It Stick tell you to do, and is it still right? The book asks you to test yourself instead of rereading, space your study sessions out, mix different problem types together, and tie new material to what you already know. Its two biggest bets, self-testing and spacing, still sit at the top of the evidence.

The rest holds up more unevenly. Some techniques rate as moderately useful. Some were never rated in the review most people cite. One of them, reflection, has strong support in classrooms, but with a condition attached that matters if you study alone.

This page summarizes the recommendations, checks each one against the evidence we can cite, and marks where that evidence stops.

Most of the book is an argument for six practices:

• Retrieval. Pull information out of memory by quizzing yourself instead of looking it up again.
• Spacing. Return to material over days or weeks instead of in one sitting.
• Interleaving. Mix different kinds of problems in one session instead of doing a block of one type.
• Elaboration. Explain new ideas in your own words and connect them to what you already know.
• Generation. Try to answer a problem before anyone shows you how.
• Reflection. Look back after studying or a test and ask what worked and what to change.

The book also rejects "learning styles", the idea that you learn best when material matches a preferred mode such as visual or auditory.

The common thread is that the study that feels easy, such as rereading a chapter, tends to produce less durable learning than the study that feels effortful.

The most useful check is Dunlosky and colleagues' 2013 review in Psychological Science in the Public Interest. It rated ten common study techniques as high, moderate or low utility, meaning how likely each is to help across learners, materials and tests. A summary of those ratings lists them without any later revision, and they line up with the book like this:

Only two of the ten techniques in the review got the top rating, and they are the book's first two. Both reinforce each other in practice: a quiz you take again three days later is retrieval and spacing at once. For what a newer medical-education meta-analysis adds on timing, see how spacing changes a schedule.

The same review rated rereading, highlighting and summarizing as low utility. That is the flip side of the book's argument, and it is why a last-minute cram feels productive and fades fast.

Interleaving and elaboration landing at moderate does not mean they fail. It means the review found them helpful under narrower conditions, or on thinner evidence, than self-testing and spacing.

Reflection is the part of the book closest to what educators call metacognition: planning how you will learn, checking whether it is working, and evaluating afterward.

The evidence here is strong, but it comes from classrooms. England's Education Endowment Foundation, the government-designated "What Works Centre" for education, rates metacognition and self-regulation as low cost with 4 out of 4 for evidence security. As of September 24, 2026, its Toolkit page listed an impact of +7 months of additional progress. The refreshed guidance report, drawing on more than 355 studies per secondary reporting, is reported to put the figure at about +8 months for disadvantaged pupils.

The catch is in the same report:

The benefit shows up when reflection is built into a specific subject, and the guidance stresses that planning looks different in art than in math. Those figures also describe teacher-led school instruction, not an adult working through a book alone. One summary of the Toolkit notes that effect sizes are averages, not guaranteed outcomes.

For you, that means reflecting on your organic chemistry mistakes will likely help more than reflecting on "how you learn" in general. A running record of wrong answers is one concrete form of this, as in an error log for a retake.

Three claims deserve a check that the sources behind this page cannot give.

Generation was not among the techniques the 2013 review rated, so this page has no utility rating to report for it.

The rejection of learning styles is not tested by any source cited here. We are not endorsing or disputing it on this page.

The boundary conditions on interleaving and retrieval, such as how much prior knowledge interleaving needs or how well lab results carry into real courses, come from later meta-analyses that are outside what we can cite on this page. The moderate rating for interleaving is a signal that such limits exist. It does not tell you what they are.

What would settle these points is direct evidence on self-directed adult learners studying real course material, which is the situation most readers of the book are in.

Yes, if you read it with the ratings in hand. Build your week around self-testing and spacing, because that is where the book and the evidence agree most strongly. Treat interleaving and elaboration as useful additions, not foundations. Keep reflection tied to the subject you are studying.

Where the book states something with more confidence than the research, as with generation, hold it loosely until you see the study behind it.