How to Improve Productivity at Work by Reading Less

How to improve productivity at work, in one sentence: stop counting what you finish and start counting what you let in. Most of the lost hours are not slow work. They are unplanned reading, half-answered messages and tabs opened "for later". Cut the intake, set one block of protected time, and the output follows.

That is a smaller claim than most advice makes. It is also one you can test by Friday.

Where does the working day actually go?

I kept a log for two weeks last spring. Every time I switched tasks, I wrote one line. By day four the pattern was embarrassing: the work itself was fine. The switching was the problem.

The best-known research on this comes from Gloria Mark at UC Irvine. Her team observed office workers and found that people took roughly 23 minutes to return to a task after an interruption, as summarised in this Fast Company piece on the cost of task switching. Treat the exact figure loosely. The original study reports wide variation, and short interruptions inside the same project cost far less than a jump to something unrelated.

What survives the caveats is the shape of the problem. A switch is cheap to start and expensive to recover from. Roughly half of those switches, in her data, were self-inflicted.

That second point matters. If half the interruptions come from you, a better inbox rule will not fix the day.

Why does a longer to-do list make things worse?

A to-do list is a queue with no back pressure. Anything can be added, nothing has to leave, and every item you see again costs a small decision.

I used to run a list of about sixty items. I looked at it each morning, felt busy, and picked whatever was easiest. The hard, important item sat at position forty-one for three weeks.

The fix was not a better app. It was a shorter list with a rule: three items for the day, written the night before, one of them uncomfortable. Everything else goes into a parking note that I read once a week and mostly delete.

Skip any system that asks you to categorise every item by colour. The categorising becomes the work.

What does a protected block really look like?

"Deep work" has become a slogan, so here is the unglamorous version. One block, ninety minutes, same time each day, with the notification centre closed and the phone in another room.

It does not need to be the morning. Mine is mid-morning because that is when my team is quiet. Pick the hours when you are least likely to be needed, not the hours you imagine you are sharpest.

Three details make or break it:

• Tell one person where you are, so silence does not read as absence.

• Write the single outcome on an index card before you start, so the block has an end.

• Keep a short list of "things to look up later" beside you, so a stray thought does not become a browser tab.

That last one is the cheapest trick I know. A thought written down is a thought you can stop holding.

Is reading the hidden cost nobody tracks?

Ask people what eats their week and they say meetings. Fair. But there is a second pile that rarely gets measured: the reading that arrives uninvited. Industry newsletters, long threads, a colleague's "worth a look" link, a vendor report that landed in three channels.

None of it is urgent. All of it feels slightly obligatory. So it stays open, and the open tabs follow you into every block you try to protect.

My own rule is blunt. Anything I save to read gets a decision within a week: read it, summarise it, or delete it. The queue is allowed to be short. It is not allowed to be a museum.

A summary is often enough. If a forty-page report matters to you, the first useful question is whether it matters to you in 150 words. Usually the answer tells you whether to read the rest. If you want a tool for that step, a read-it-later app with decent highlighting does most of the work.

Be honest about the limit, though. A summary tells you what a piece says. It cannot tell you whether you believe it. For anything you will act on, read the original.

Which meetings can you replace with a written note?

I am not against meetings. I am against meetings whose only output is that people now know the same thing. That is a document, not a meeting.

A test I use: if the agenda has a verb like "update" or "share" and no verb like "decide", it can probably be a written note with comments open for a day. Plenty of teams already work this way, quietly.

When the meeting does happen, the useful artefact is the record. Someone should be able to skip it and still know what was decided, who owns what, and by when. Recording and transcribing helps, as long as somebody actually reads the output, which is the part people forget.

If your meetings are mostly external calls, a plain transcript tool is enough. You do not need a platform; you need a searchable text file you will open on Thursday.

A word of caution. Transcripts create their own reading pile. If you record everything and read none of it, you have swapped one backlog for another. Decide in advance which meetings earn a transcript and which only need three lines of notes.

How much should you rely on AI assistants?

Used narrowly, they remove friction from tasks you already understand: drafting a first version of a routine email, restructuring rough notes, pulling action items from a long thread. Used broadly, they produce a lot of competent-sounding text that someone still has to check.

My working rule is to use an assistant where a wrong answer is cheap to spot and expensive only in time. Reformatting a document: yes. Summarising a contract you must sign: read the contract.

Whichever tool you pick, measure it for two weeks. If you cannot say what it saved in minutes, it probably saved nothing and cost a subscription.

How do you handle email and chat without living in them?

Treat them as post, not as conversation. Three check times a day, say ten o'clock, one and half past four, each with a ten-minute cap. Anything that needs more than two minutes becomes a line on the parking note, not an immediate reply.

People worry that this reads as slow. In my experience it reads as reliable. A colleague who answers in three hours every time is easier to plan around than one who answers in three minutes on Monday and three days on Thursday.

Set expectations once, in writing. A single line in your status or signature, something like "I check messages at 10, 1 and 4:30; call if it cannot wait", removes most of the guilt. It also reveals how few things truly cannot wait.

What does a realistic week look like?

Here is mine, trimmed. Monday morning, thirty minutes to empty the parking note and set the week's single outcome. Daily, three items written the night before and one ninety-minute block. Wednesday, a decision pass over saved reading: read, summarise or delete. Friday afternoon, five minutes to count finished items and leftover tabs.

That is under two hours of overhead across the week, and most of it replaces time I used to spend feeling vaguely behind.

It will not survive a crisis week, and it should not try to. When a launch or an outage takes over, drop the rituals and keep only the three-item list. Pick the system back up the following Monday without a lecture to yourself.

Expect the first fortnight to feel slower. You are removing the small hits of progress that come from clearing easy items, and the replacement is harder work done less often. That discomfort is the sign it is doing something.

What should you skip?

A short list, because opinions are cheaper than systems.

• Skip elaborate planning rituals that take longer than the plan they produce.

• Skip the always-on chat window. Check it at set times, like post.

• Skip tools that promise to organise everything automatically. You will still have to organise the organiser.

• Skip checking email first thing, unless your job is email.

• Skip comparing your output with someone's highlight reel of morning routines.

None of this is austere. It is just fewer decisions per hour.

How do you know it is working?

Do not measure hours worked. Measure two things: how many of the three daily items you finished, and how many tabs and unread items you carry into Friday.

Check them on a Friday afternoon, in five minutes, with a plain notebook. If the first number rises and the second falls over three weeks, the change is real. If neither moves, change one thing only, the block time or the list length, and run it again.

I still slip. Last month I let the parking note grow back to forty items and spent a week feeling behind for no good reason. Deleting thirty of them took ten minutes.

Better productivity at work is mostly a question of what you allow in. Start there, and keep the habit smaller than the ambition.