You didn't run out of time because the section is too long. You ran out because three or four questions each took four minutes, and the easy questions at the end got whatever panic was left over. That's not a knowledge problem — it's a spending problem, and it has a mechanical fix.
The math you're up against
The digital SAT gives you 32 minutes for 27 Reading & Writing questions per module. That's about 71 seconds per question — on average. But the questions aren't average: a punctuation question with a mechanical rule check can take twenty seconds, and a dense inference question can honestly need two minutes. A single flat pace is the wrong tool for an uneven test.
The failure mode has a name: the time-sink — one hard question that doesn't just cost its own point but mugs the two or three questions after it, which you rush with a drained clock. If you want the full anatomy of that leak, and why the damage shows up downstream disguised as careless misses, the time-sink guide covers it. This page is the fix.
The 90-Second Flag: the full protocol
Two passes through the module. That's the whole architecture.
Pass 1 — answer what clicks. Work in order. If a question is coming together — you see the rule, you have a prediction, you're closing on an answer — finish it. If you hit about ninety seconds and it still isn't clicking: pick the best answer you have right now, mark it with the flag tool, and move to the next question. Do not stay to "just check one more thing." That sentence is how four-minute questions happen.
Pass 2 — the flag round. When you finish question 27, go back to your flagged questions with the time you banked. Take them in any order — easiest-looking first is fine. They often crack on a second read.
Two rules inside the protocol:
Always guess before you flag. Never leave a blank, even for a second. The SAT has no wrong-answer penalty, so a guess is free upside and a blank is the only guaranteed zero. The guess is also insurance: if the clock dies before you get back to a flag, you're on record with your best attempt instead of nothing.
Decide the trigger before the test. You won't run a stopwatch on every question, and you don't need to. "I've read this twice and I still don't know what it's asking for" is past the line. The point of the number is that the decision is already made — you're executing a policy, not negotiating with a hard question in the moment. A hard question never feels hopeless mid-attempt; it feels almost-solved. That's exactly why the line has to be drawn in advance.
Why 90 seconds and not 71
Seventy-one seconds is the average budget, not a ceiling. Rule-based grammar questions routinely come in far under budget, so a normal Pass 1 banks time on its own. Ninety seconds is budget plus slack: high enough that you're not flagging half the section, low enough that no single question can eat the three after it. The exact number matters less than having one — a line set in advance is enforceable, and "I'll move on when it feels right" is not.
The pace checkpoint: question 14
One mid-module check, so a slow drift can't survive until minute 30. Question 14 sits at the halfway mark of 27. When you reach it, glance at the clock: you should have about half of your 32 minutes left. Ahead — change nothing. Behind — tighten the trigger: flag sooner and trust the flag round. One glance, one adjustment, done. Checking the clock every question is its own leak; it buys anxiety, not information.
Why the flag round feels different
Pass 2 is calmer than Pass 1, and there are real reasons why. The clock pressure is structurally different: every remaining minute belongs to a small set of questions you chose, not to an unknown stretch of test. The sunk cost is gone — walking away from a question after three minutes of work feels like abandoning an investment (the sunk-cost effect, a well-documented bias), but returning in Pass 2 carries no such weight, because your guess is already banked and the worst case is already handled. And a second read is a genuinely fresh read: what overloaded your working memory the first time often assembles cleanly the second.
None of this makes hard questions easy. It makes them cost what they should: one question's worth of time, one point at stake — instead of wasted points bleeding out of the questions around them. If misses like these come from habits rather than missing knowledge, the behavioral leaks guide maps the other two big ones.
The self-test: one timed module, scored three ways
Run this today. Take one timed Reading & Writing module (32 minutes) using the 90-Second Flag. When it ends, write down three numbers:
- Flags planted — questions you flagged in Pass 1.
- Flags converted — flags you turned into correct answers in Pass 2.
- Downstream misses — questions missed immediately after any long question.
Compare against your last normal section, where the long questions ran unchecked. If the protocol is doing its job, the third number drops — the easy points stop leaking — and you'll have made the case with your own data, not anyone's promise. One caveat: if the first number is huge (eight, ten flags), that isn't a pacing problem. It's a content gap wearing a pacing costume, and the plateau playbook shows you how to sort your misses into buckets and fix the real thing.