TOEIC Link Pacing and Time Management: How to Spend Your Minutes on an Adaptive Test

Pacing on TOEIC L&R is a solved problem. You have 75 minutes for 100 reading items, you know roughly how to budget Part 5 versus Part 7, and most prep books give you a stopwatch routine that works. Pacing on TOEIC Link is a different problem because the test ends when the engine has enough information about your level — not when the clock runs out, and not when you have answered a fixed number of items.

That difference reshapes which pacing instincts help and which ones hurt. Test-takers who carry over their L&R timing habits often finish their first TOEIC Link session with the right strategy for the wrong test. This guide explains where the minutes actually go on TOEIC Link, the three places where time leaks the most, and a per-module pacing playbook you can rehearse before test day.

How TOEIC Link allocates time

TOEIC Link sets a maximum module duration, but the median test-taker finishes well inside it. The engine uses Computerized Adaptive Testing to terminate as soon as it has confirmed your CEFR level with the required precision. For a typical test-taker, this means:

• Listening: 25-30 minutes (max 35)
• Reading: 30-35 minutes (max 45)
• Speaking: 12-15 minutes (max 20)
• Writing: 18-22 minutes (max 30)

The maximums matter only for borderline test-takers — those whose ability sits exactly on a CEFR boundary. The engine needs more items to confirm whether you are at B1 or B2, and you may use the full time window. For everyone else, the test ends faster than the maximum suggests.

This produces a pacing paradox. On a fixed-form test, slower pacing equals fewer items answered, which equals a lower score. On an adaptive test, the relationship is non-linear. Spending 90 seconds on a single item still gives the engine information about your ability, but it costs you items in the rest of the module. There is no simple "answer faster, score higher" rule.

Where minutes actually leak

In our review of test-taker self-reports and proctoring logs, three patterns dominate the time leakage problem.

Leak 1: Over-investment in the routing phase

The first 5-8 items of each module are calibration items drawn from a medium-difficulty band. They look manageable, and test-takers often spend extra time on them trying to "start strong." This is misallocated effort. The routing phase is intentionally noisy — the engine knows it does not have enough information yet to pick items at your real level, and it weights these items lightly when calculating your final estimate.

A test-taker who spends 90 seconds on each routing item burns 7-12 minutes before reaching the items that actually drive the score precision. Those minutes are gone, and the items that come next — the calibration phase items, where the engine is actually narrowing in on your level — get less time as a result.

Leak 2: Sunk-cost re-checking

On an adaptive test, you cannot return to previous items. The interface enforces forward-only progression. Despite this, test-takers spend mental cycles re-checking items they have already submitted, which slows down their work on the current item without changing any answer. This is a behavioral leak, not an interface leak — the test-taker knows they cannot go back, but the habit persists from L&R.

The fix is to commit to the current item with a hard rule: once submitted, the item is closed and you do not think about it again. This is harder than it sounds for test-takers who are used to flagging questions for later review.

Leak 3: Termination misinterpretation

The test ends when the engine confirms your level. Some test-takers, used to fixed-form tests where every minute of the time window is information-bearing, assume an early termination is bad news. They slow down or second-guess their answers near what they sense is the end, hoping to extend the test and "improve" their score. The engine does not work this way — extending the test by deliberately answering more slowly does not improve precision, and may push the engine toward a more cautious lower estimate as it reads the slow answers as uncertainty.

The right interpretation is the opposite: an early termination usually means the engine confirmed your level cleanly, which is a good sign about your performance, not a bad one.

Per-module pacing playbook

The four modules each have different optimal pacing, partly because of the item types and partly because of how the engine calibrates each module separately.

Listening module

Audio items have a fixed playback length, which means a portion of your time is non-controllable. You cannot pause or replay (in standard delivery), so the only variable you control is response time after the audio ends.

• Routing phase: Aim for 15-20 seconds response after audio. The questions are not designed to be tricky here.
• Calibration phase: 20-30 seconds is the sweet spot. The engine is selecting items right at your boundary, and the difficulty is real. A few extra seconds of careful processing pays off.
• Precision phase: Stay disciplined. The items here are at the cut score for your final level — do not over-think, and do not rush.

A common Listening pacing mistake is replaying the audio mentally during the response window. This eats response time and rarely changes the answer. If you did not hear it the first time, your second mental pass usually will not catch it either.

Reading module

The reading module is where most pacing variance shows up because reading items range from 30-second word-meaning items to 4-minute long-passage items. The engine mixes these as it calibrates.

• Routing phase: 45-60 seconds per item is appropriate. Easy passages should be quick.
• Calibration phase: Highly variable. A long passage at your ability boundary may legitimately need 3-4 minutes. Do not panic.
• Precision phase: Same as calibration — accept the time the item demands.

The reading module is where the reading module guide becomes most useful for understanding which item types deserve the time investment.

Speaking module

Speaking is heavily time-controlled by the prompt structure. You have a preparation window (15-30 seconds) and a response window (30-60 seconds depending on item type). Pacing here means using the preparation window deliberately.

• Use the full preparation window. Test-takers who start speaking immediately almost always score lower than those who use the full window to plan structure.
• Do not pad the response window. If you finish your answer before the timer ends, stop. Padding with filler ("uh, you know, basically") signals lower fluency to the rater.

Writing module

Writing has explicit time limits per item (30 seconds for sentence-level items, 5-8 minutes for paragraph items, longer for the essay). Pacing here means hitting the structure inside the limit, not maximizing word count.

• Sentence items: 20-25 seconds for sentence construction, 5-10 seconds for review.
• Paragraph items: 60-90 seconds outlining (mental or scratch), then writing.
• Essay item: 5 minutes outlining, 18-20 minutes writing, 3-5 minutes review.

The speaking and writing tips guide covers item-specific structures in more detail.

The "first pass complete" mental model

The most useful pacing mental model we have seen is the "first pass complete" frame. On a fixed-form test, you have time for multiple passes — first answer everything, then re-check, then guess on remainders. On TOEIC Link, you only have a first pass. The interface does not allow second passes, and the engine reads each item the first time you submit.

This means your default response on every item should be the answer you would give if you knew you could never come back. If you would normally flag an item for review, change the rule: answer it now, with the best confidence you have, and move on. The pace this produces is faster than test-takers usually allow themselves, but it is the right pace for the test format.

Practicing pacing under realistic conditions

Pacing is hard to practice without the actual interface, because the time pressure of a fixed-form practice test does not match the engine-driven progression of TOEIC Link. The closest substitutes:

• Time yourself per item, not per section. Aim for the per-module averages above.
• Use a forward-only practice mode (some prep platforms offer this) where you cannot return to previous items.
• After each practice session, log items where you spent more than 90 seconds. Look for patterns — if 80% of your over-time items are inference reading questions, that is where to focus prep.

A pacing rehearsal protocol that has worked for our test-takers: three sessions per week, one focused on each pace component (per-item timing, forward-only commitment, termination acceptance), with a fourth session integrating all three. Two weeks of this protocol typically reshapes the pacing instincts that L&R installed.

Bottom line

TOEIC Link pacing rewards calibrated effort across the test, not heroic effort on individual items. The test ends when the engine has enough information — your job is to give it clean signal as efficiently as you can, then accept the termination when it comes. Pacing well on TOEIC Link is partly a skills problem and partly an unlearning problem: the L&R habits that earned you a clean 900+ may be exactly the habits eating your TOEIC Link minutes.

For test-takers transitioning from TOEIC L&R, the transition guide covers the broader habit shifts. For day-of-test routine, the test day checklist integrates pacing prep into the broader prep flow.