You've written out your answer to "tell me about a time you handled conflict." You've read it three times. It sounds good. You feel ready.
Then the interviewer asks it slightly differently than you rehearsed, and the answer you felt so ready with doesn't come out. Not because you didn't know it, you did, on the page. It just wasn't stored the way you needed it to be, as something you could produce from nothing, out loud, on the spot.
Recognition feels like readiness. It isn't
There's a well-documented gap between recognizing information and recalling it. Reading your own answer back to yourself is recognition: you see the words, they seem right, your brain nods along. That's a real feeling, but it's not the skill an interview actually tests.
An interview tests recall under mild pressure: someone asks a question, and you have to pull a coherent answer out of memory, structure it as you go, and say it out loud, in real time, with no text in front of you to lean on. That's a different task, and being good at one doesn't automatically make you good at the other.
This is why a candidate can walk out of a prep session feeling completely ready, then freeze on a question they'd technically "prepared," word for word, the night before.
Why re-reading notes feels so productive
Re-reading is easy, which is exactly why it's appealing the night before an interview. There's no friction, no chance of getting it wrong, no awkward pause while you search for the next sentence. It feels like progress because it's smooth.
Retrieval is the opposite. It's uncomfortable by design, you're pulling something out of memory with no guarantee it'll come out clean, and that discomfort is often mistaken for a sign you're not ready, rather than a sign you're doing the actual work that builds readiness.
What retrieval practice actually looks like
In practice, it's simple: someone (or something) asks you the question, out loud, and you answer out loud, with nothing written in front of you. No script, no bullet points to glance at. If you can't remember exactly how you phrased it last time, that's fine, that's the point. You're rebuilding the answer from the underlying facts each time, not replaying a memorized script.
The distinction matters because interviews rarely ask the exact question you rehearsed. They ask a slightly different version of it, or a follow-up that assumes you already covered the first part. A memorized script breaks the moment the question doesn't match it word for word. An answer you've actually retrieved and rebuilt a few times flexes to fit the real question, because you're working from the substance, not the script.
Adding pressure is what makes it transfer
There's one more piece that pure retrieval practice on your own often misses: mild pressure. Practicing an answer alone, at your own pace, with no one listening, is retrieval, but it's retrieval without the part that actually derails people, someone watching, a slightly awkward pause, a follow-up question you didn't expect.
That's the gap between "I can say this answer well in my room" and "I can say this answer well when someone's evaluating it." Closing it takes practice that includes both parts: producing the answer from memory, and doing it under conditions that resemble the real thing, ideally with something that pushes back when the answer is thin instead of letting it slide.
A better way to spend your last few days
If you have a stack of written-out answers sitting in a doc, don't throw them away, they're a good foundation. But the night before an interview is better spent saying them out loud from memory, getting them wrong, fixing them, and saying them again, than reading them one more time.
A rough rule of thumb: if you've only ever seen your answer on a screen, you haven't practiced it yet. You've reviewed it. The practice starts the first time you say it out loud with nothing to read from.