The document sent before the meeting: read it, skim it, or turn up cold

Should you read the pre-read, or let them talk you through it?

Eleven slides in your inbox at twenty to five, meeting at nine, and "please review beforehand" in the body of the mail. The question is not whether you have twenty minutes. It is whether reading the thing yourself buys you anything that sitting there while somebody walks the deck does not.

Clinton-Lisell pulled that comparison together in Review of Educational Research in 2022, pooling 46 studies and 4,687 readers of various ages. Taken as a whole, reading and listening came out level. The overall difference was g = 0.07 and did not reach significance, which is the honest headline and not the useful part.

The useful part sits underneath it. When readers set their own pace, reading came out ahead at g = 0.13. When somebody else set the pace, the reading advantage vanished, and the estimate pointed the other way at g = -0.32, though that one was not reliable either. What separated the two conditions was who controlled the speed, not which sense was doing the work.

Then the second split, which is the one that decides whether a pre-read earns your twenty minutes. On inferential comprehension, the kind where you have to connect two things the text never connects for you, reading beat listening at g = 0.36. On general comprehension the gap was 0.15. On literal recall, where you just need the number from slide four, it was g = -0.01.

Nothing at all. That is the part most people read hardest.

So the shape of the question waiting for you in the meeting decides whether to read. If you are going in to approve a figure somebody will read out, listening is fine and the research says so. If you are going in to say whether the plan holds together, whether the assumptions survive contact with your team, or what breaks in month three, that is the inferential question. It is the one place where reading at your own pace has a measured edge. Read for the judgment, not the recall.

Keep the size of that edge in view. A g of 0.13 is small by any convention, the overall comparison found no reliable difference at all, and none of those 46 studies put anybody in a meeting. They tested readers on comprehension passages, which is a cleaner object than a deck with three authors and the real ask buried on page nine. Carry the pace finding across to your own week. Do not carry any impression that somebody has measured this with your Tuesday in mind. Nobody has.

Can you follow the document while someone presents it?

Reading along while they present looks like the efficient compromise, and it is the one arrangement anybody has tested head to head against your other options.

Hui and Godfroid ran it as a registered report in Language Learning, which means the design and the predictions went on the record before any data came in. Eighty-six intermediate to advanced Chinese learners of English took in excerpts of a novel three ways: reading silently, listening only, and reading while the audio played. The authors expected the audio to take load off decoding and free up room for comprehension, and they said so in advance.

It went the other way. Reading while listening produced worse comprehension than reading silently. Both reading conditions beat listening on its own. A preregistered prediction coming out backwards is the version of a result you can lean on hardest, because nobody went looking for it after the fact. Your compromise is the condition that lost.

Hold the obvious caveat alongside it. A second-language reader following a novel is not you following a slide, and the decoding load the study was built around is much heavier in a language learned as an adult than in your own. The mechanism may simply not be the one running in your meeting. What survives the trip is narrower: doing both at once is not free, and no one has shown it beats reading on its own.

So pick one.

If the document is going to be narrated anyway, read it before or read it after, and spend the narration listening for what the presenter adds that the slides do not say. Half-following the deck while they talk is the only way of taking it in that has been tested directly, and it came out behind the way you would have done it on your own.

There is a version of this worth asking for. Plenty of pre-reads exist so that nobody has to present them, and those meetings open with ten silent minutes to read the thing in the room. That hands you the self-paced condition, just late. When the alternative is a deck narrated at the sender's speed, it is a reasonable thing for you to propose to whoever called the meeting.

What do you do when the deck arrives an hour before?

An hour is not a read. It is a triage, and the evidence above tells you which part to spend it on.

Work backwards from the question you will be asked. The reading advantage lives in inferential comprehension, so the pages worth your own eyes are the recommendation, the assumptions holding it up, and whatever the numbers are supposed to prove. Background and context are what you can afford to be told, since that is where reading and listening came out level. This is the Pareto Principle pointed at somebody else's writing, and it is uncomfortable to apply honestly, because the pages that take longest to read are rarely the ones carrying the ask. Skim the setup. Read the ask.

Give the read a slot rather than an intention. Timeboxing fifteen minutes at a named time is what stops the hour draining into whatever is already open on your screen, and fifteen deliberate minutes on three pages beats an hour of meaning to get to all eleven. Name the time now.

One more finding belongs here, before you decide you have read it. Clinton's 2019 meta-analysis of paper against screens covered 33 randomised experiments and found that readers judged their own comprehension more accurately on paper than on screen, at g = 0.20. Comprehension itself was lower on screen at g = -0.25, concentrated in expository text, which is what a work document is. Which medium to choose is a focus question with its own page; the half that matters here is the calibration one. A skim on screen feels more like a read than it was.

Which is how people arrive at nine confident and empty.

The reverse case deserves a word, since in practice it is the more common one. You read it, and five minutes in it is obvious that most of the room did not. Resist summarising the document back at them. You are the one holding the inferential advantage the research describes, and the way to spend it is on the question nobody else is equipped to ask yet, not on a recap the author could give faster.

When the hour ran out and you did not get there, say so in the first minute rather than letting it surface in a question you cannot answer. One line covers it: you have not read it, here is the ask as you understand it, and you will come back on anything that turns on the detail. The room finding out slowly costs you more than the sentence does. Deciding that line before the meeting rather than during it is the same work as preparing anything else you plan to say, and there is a page here on doing exactly that.

How does your working style change the answer?

Which part of this you get wrong tends to follow how you already work. The four names below are this site's own, from the working-style self-check on the home page.

Architect-leaning readers read all eleven slides and annotate them, which is the version with the least left over. Reading showed no advantage at all on literal recall, so the pages you worked hardest on are the ones the meeting would have given you for free. Read the recommendation and its assumptions closely, and let the background be narrated at you.

Sprinters read it in the four minutes before the call connects, which is the self-paced condition in name only. You are reading at the speed the calendar set, not the speed the argument needs. Booking the fifteen minutes the afternoon before is the whole fix, and it works precisely because the slot sits somewhere other than the gap in front of the meeting.

Visionary-leaning readers skip the document and arrive with the strategic question, which is often the right question to ask. It is also, often enough, the one the deck answered on page six. Four minutes on the ask and the assumptions keeps your question from landing on ground the author already covered.

Improvisers turn up cold on purpose and do well in the room, because responding to what is actually said is the thing you are good at. The research does not take that away: overall, listening came out level with reading. Where it costs you is the inferential question, the moment somebody asks whether the plan holds, and you are working it out live from a document everybody else has had since Thursday. Cold works until it does not.

None of that is a reading-speed problem.

So the next move is the same for all four, sized differently. When the next pre-read lands, decide which question the meeting will put to you, read only the part that answers it, and read it before anybody starts talking. If no question will be put to you at all, the document is not the problem and the meetings pillar covers what is.

Frequently asked questions

Sources

• Clinton-Lisell (2022), Review of Educational Research: Listening ears or reading eyes - a meta-analysis of reading and listening comprehension comparisons
• Hui and Godfroid (2026), Language Learning: Listening, reading, or both? Rethinking the comprehension benefits of reading-while-listening
• Clinton (2019), Journal of Research in Reading: Reading from paper compared to screens - a systematic review and meta-analysis