How to Autopsy a Practice Test

A practice test you don't review is expensive anxiety with a number stapled to it. You gave up a Saturday morning, sat the full three hours, felt the nerves, and at the end you bought yourself a single data point: 1380, or 1290, or whatever it was. If that number is all you carry away, you paid full price for the test and collected almost none of what it was worth. The points you are hunting are not in the score. They are in the forty minutes after it.

That review has a real procedure. Not the vague instruction to "go over your mistakes," but a specific sequence you can run on every test you take. Here it is, step by step.

Run it the same day, while the test is still warm

Do the autopsy the same day you take the test, ideally within a few hours. This is the step you are most tempted to skip, and skipping it quietly wrecks everything that comes after.

The reason is simple. Right now, an hour after the test, you can still reconstruct what you were thinking on question 17: you read it as a main-idea question, you liked B because it sounded the most important, you never registered the word "except." Wait a week and that memory is gone. You will look at question 17, see that you picked B, see that the answer was D, and have no idea why B tempted you. And the why is the entire point. Without it you are left with "I got it wrong," which you already knew and which fixes nothing. A practice test is a crime scene. You want to walk it while the evidence is still on the floor.

For every miss, name the real cause in one sentence

Go through those questions one at a time. For each, write a single sentence naming what actually went wrong. One sentence, in plain language, and it has to be mechanical rather than moral. "Careless" is banned. So are "silly," "dumb," and "I knew that." Those are verdicts on yourself, and they explain nothing. You cannot practice your way out of "careless," because it doesn't name a thing you did.

What you are after is the specific failure, and it almost always lands on one of five:

• You misread what the question asked. You answered "support" when it said "weaken," solved for x when it wanted x plus 3, found the main idea when it asked for one detail.
• You didn't know the thing. A vocabulary word, a grammar rule, a math concept. An honest knowledge gap, which is the most fixable cause on this list.
• You fell for the trap answer. The choice that was true but didn't answer the question, or true but went a step further than the passage did. You saw the bait and took it.
• You ran out of time. You rushed the last four questions and missed things you'd have caught with thirty more seconds, or never reached them at all.
• You talked yourself out of it. You had the right answer selected, second-guessed under pressure, and changed it to a wrong one.

The naming is the work. The companion post on this blog about the "silly mistakes" that aren't silly calls this one sentence the revelation, and it is: the moment you write "I picked the answer that restated the passage instead of the one that challenged it," you have converted a shapeless bad feeling into a specific, drillable fact.

Do the same for the ones you got right by luck

This is the step you will want to skip, and it is worth more than it looks. When you narrowed it to two choices and guessed, or the answer was right but you couldn't have said why, mark it and reconstruct it exactly like a miss. A right answer you didn't earn is a wrong answer on a delay. The coin came up heads today. It will land tails on the real test, on a question that counts, and you'll have no idea why, because you never looked at it. Every guessed-right question is a miss the scoreboard happened to hide from you. The autopsy is where you go find them before the test does.

Sort the causes into three or four piles, not twenty

Once you have a stack of one-sentence causes, ten or fifteen of them, do the thing that turns a pile of mistakes into a study plan: group them. Put the sentences that share a cause next to each other. What you'll find, when you do this honestly, is that you did not make fifteen different mistakes. You made three or four mistakes, several times each.

Nine times out of ten, the fifteen sentences collapse into three or four. Four of them are really "I misread the constraint word in the question." Three are "I took the trap that goes one step too far." Two are "I rushed the end of the module." That collapse is the most valuable thing the autopsy produces, because it tells you that you don't have fifteen problems to fix. You have three, and each one keeps showing up.

Frequency is the signal. A cause that appears once is probably noise, the equivalent of one stray question on your score report's skill bars. A cause that shows up five times across a single test is not noise. It's a habit, and it is precisely the thing pinning your score where it is. If you've taken a few tests and the number refuses to move, this pile is why: the post on why your score is stuck is about this same repeat offender, seen from the outside as a plateau.

Turn each pile into a rule you can actually run

For each pile, write one instruction for your future self. This is where a lot of review dies, because the instruction that comes naturally is "be more careful," and "be more careful" is not a rule. It's a wish. It gives your brain nothing to do differently on the next question; it just asks you to feel more tense while you do the same thing.

A real rule is specific, and it fires at a specific moment. It names the trigger (the kind of question) and the action (the exact move to make before you answer). Compare:

• Wish: "Stop rushing transition questions."
• Rule: "On a transition question, before I look at a single answer choice, I say the relationship between the two sentences in plain words: this one contradicts the last, or gives an example of it, or states its result. Then I find the transition that matches the relationship I already named."

The second one is executable. It tells you what to do, when to do it, and it does its work at the point where the mistake actually happens, which is before the answer choices get their hands on you. Write one of those for each pile. Three or four rules per test, phrased concretely enough that you could hand them to a stranger and they'd know exactly what to do.

The fix isn't real until you re-test it

Here is the part that separates people who review from people who improve. Writing the rule feels like finishing. It's the halfway mark. Nodding at a mistake, understanding it, even writing an elegant sentence about it, does nothing on its own. You have recognized the problem. Recognition and repair are different events.

So schedule the re-test. A week or two after the autopsy, pull a fresh set of questions of the exact type your rule targets. Fresh is the operative word: not the same questions you already saw, because re-doing those tests your memory of the answer, not your skill at the question. New questions, cold, with the rule in hand. Run it. If you still miss them, the rule didn't take, and you rewrite it or drill it harder. If you clear them without strain, now you've earned the right to call it fixed. Not a day before.

This is the difference between two skills that are easy to confuse: recognizing the right answer when it's shown to you, and producing it yourself, under time, with the trap sitting right there. The autopsy plus the spaced re-test is the only honest way to move from the first to the second.

Watch the film like a coach, not the player

One thing decides whether you can do any of this, and it isn't intelligence or discipline. It's the stance you take toward your own mistakes.

There are two ways to watch the tape of a game you lost. The player watches it and relives the loss: winces at the fumble, feels the old embarrassment, wants to look away. The coach watches the same footage cold. Freezes the frame on the blown coverage, says "there, that read was late," and writes tomorrow's drill. Same tape, opposite use. One is emotion. The other is information.

Your autopsy has to be the coach's screening. The moment it curdles into "I'm so stupid, how did I miss that," it stops working, because a person who feels ashamed of a mistake wants the mistake out of sight, and this whole procedure requires you to hold each one still and stare at it. Cold, specific, unsentimental. A wrong answer is not a referendum on you. It's data about a process, and a process is adjustable in a way that a verdict on your character is not. You are not the player here. You are the analyst in the booth, and the game is already over: nothing left to feel about it, plenty left to learn from.

That is the whole method: same-day, one sentence per miss and per lucky guess, sort into three or four piles, one runnable rule each, re-test on fresh questions later, all of it in the coach's voice. Run it after every practice test and each test stops being a number and becomes a set of instructions for the next two weeks. It works. It also takes a real hour of unglamorous labor and a certain ruthlessness about your own reasoning that is genuinely hard to keep pointed at yourself, test after test.

That labor is the exact thing Forge was built to take off your hands. It watches how you actually move through a full diagnostic (where you slow down, where you second-guess, which questions you misframe) and hands back the piles already sorted: the real causes and the specific thing to drill. The autopsy done for you, without the hour and without having to be your own coldest critic. Knowing what you got wrong was always the easy part.