On this page10 sections
Where did the day go? You answered an email, rang the garage, went back to the report, paid a bill, checked a delivery, then went back to the report again. Every job was small, but each one pulled you away from the last, and each time you came back you had to find your place.
Task batching offers a way to group those small jobs. Collect the ones that need the same thing, like your phone, your inbox or a trip out, and do them together in one session. The aim is fewer jumps between kinds of work, leaving longer stretches for other tasks.
Task batching at a glance
- What it is: grouping similar small jobs and doing them together in one session, instead of scattering them through the day.
- What counts as similar: jobs that need the same tool, place or person. Calls, replies, errands, paperwork.
- The bigger version: theme days, where a whole day or afternoon gets one kind of work.
- The move: small jobs wait for their batch. You write them down as they turn up, then clear them together.
- Who it helps: anyone whose day gets chopped into pieces by small jobs.
Where it comes from
The idea appears in several planning approaches. Two well-known examples show how it works.
In Getting Things Done, David Allen sorts next actions by context: the tool or place they need, such as Calls, Computer, Errands and Home. His reason is that most actions can only be done somewhere specific, so a list sorted by context lets you "see your eight calls" together and work through them.
In his 2009 essay "Maker's Schedule, Manager's Schedule", Paul Graham describes holding office hours: he grouped his meetings into set chunks at the end of his working day, so they didn't break up the time he needed for longer work.
There's lab research behind the feeling, too. In experiments by Rubinstein, Meyer and Evans (2001), people were slower when they switched between tasks than when they repeated the same one, and the extra time grew when the tasks' rules were more complex. Those were short, controlled lab tasks, so they don't tell you how much time your own day loses. But they fit what batching tries to do: switch less often.
How to batch your tasks
- Catch the small jobs. When a small job turns up, write it down instead of doing it there and then. One list is fine to start.
- Group them by what they need. Sort the list into batches: calls, replies, things to do online, things to do when you're out, paperwork. If two jobs need the same tool, place or person, they belong together.
- Give each batch a slot. Pick when it happens, like replies at 11:00 and 16:00, or errands on the way home on Thursday. A batch with no slot is just another list.
- Work the batch from top to bottom. Stay in the batch until it's empty or the slot ends. Don't start the report halfway through your calls.
- Send newcomers to the next batch. A new small job that turns up mid-session goes on the list for its next slot, unless it really can't wait.
That's the method. The skill is in choosing good batches and keeping them small enough to finish.
Theme days: batching on a bigger scale
A theme day gives a whole day, or half a day, one kind of work. Monday afternoon might be admin, Wednesday morning might be for writing, and Friday might be for errands and the house.
It's the same idea at a larger size. Instead of grouping ten minutes of calls, you group a week's worth of one kind of work. It helps when your work splits into clear types, and when you have some say over when each type happens.
It's a worse fit when jobs can't wait. If something needs doing today, don't park it until Thursday because Thursday is "admin day". A theme says what the day is mostly for. It isn't a rule that nothing else is allowed.
A worked example
Here's a made-up example. Sam works from home. By Tuesday lunchtime, these small jobs have come up:
- Ring the dentist to move an appointment
- Reply to the school about a trip form
- Renew the parking permit online
- Ring the plumber for a quote
- Post a parcel back
- Reply to two work emails that need a yes or no
- Pick up a prescription
- Pay the window cleaner
Before, Sam did each one as it came up, and the afternoon's main piece of work kept stopping and starting. So Sam sorts the list into three batches:
- Calls (2): the dentist and the plumber, at 13:30.
- Online (5): the school reply, two work replies, the parking permit and the window cleaner, at 16:00.
- Out (2): the parcel and the prescription, on the walk at 17:30.
The main piece of work gets the time between 14:00 and 16:00 without small jobs breaking it up. When an email arrives at 14:20 asking for a quick yes, Sam adds it to the 16:00 batch rather than answering it there and then.
What not to batch
- Urgent jobs. If it can't wait for the batch, do it now. Batching is for jobs that can wait a few hours or a few days.
- Two-minute jobs you're already holding. If you're sorting your inbox and a reply takes under two minutes, the two-minute rule says do it now. Batching is for jobs that need a different tool, place or moment.
- Long, deep work. Writing a report isn't a batch of small jobs. Give it its own time, and let batching keep the small jobs out of it.
Common mistakes
- Batches that are too big. A "life admin" batch with forty jobs never ends. Keep each batch small enough to finish in one session.
- Batches with no slot. If you don't decide when it happens, a batch is just another list.
- Mixed batches. A batch of a call, an online form and a trip to the shop isn't a batch. Group by what the jobs need.
- Checking anyway. Batching your replies doesn't help if you still read every message the moment it lands. Notifications can wait for the batch too.
- Waiting too long. Some jobs only take minutes but block someone else. Give replies two slots a day rather than one a week.
How to batch tasks in dypt
dypt can keep each batch in one place and bring it back on time. For now, reserve the session in your calendar app. Planning tasks as time blocks in dypt's own calendar is on our roadmap.
- Tag each job with its batch. Make tags such as
calls,online,errandsandadmin, and add one to each small job as you capture it. You can add tags as you create a task by typingtag:and the name. - Filter by the tag when the batch starts. Open the all-tasks view and filter by
callsto see every call in one list, wherever those tasks sit in your projects. Filters stay in the page address, so you can bookmark each batch. - Make the session a recurring task. Create a task like "Admin hour", set its deadline for next Monday and add a reminder. Set it to repeat weekly with its deadline carried forward, so the next session comes back at the right time. The recurring task page explains the options.
- Keep the work visible. If you use personal kanban, make the session one card with its jobs as subtasks, then do those jobs one at a time. The batch is a container, not permission to start eight things.
See custom tags for more on creating and filtering tags, and our GTD guide for using tags as contexts.
Try it this week
Pick one kind of small job that keeps interrupting you, such as calls or replies. For one week, write those jobs down as they turn up and clear them together in one or two set slots a day.
If you'd like your batches tagged, filtered and brought back each week, try it in dypt, free to start.
Next in our Do the thing videos: Getting Things Done, a whole system for capturing and organising everything you need to do.
Sources and further reading
- David Allen, "Why sorting your lists by contexts even matters" (Getting Things Done, 2010)
Context lists such as Calls, Computer and Errands, and why sorting by context helps. - Paul Graham, "Maker's Schedule, Manager's Schedule" (2009)
Grouping meetings into office hours so they don't break up longer work. - Rubinstein, Meyer and Evans, "Executive control of cognitive processes in task switching" (Journal of Experimental Psychology: Human Perception and Performance, 2001)
Lab experiments in which switching between tasks took longer than repeating one. - The two-minute rule
When to do a small job now instead of saving it for later. - Personal kanban
Limiting how many jobs are in progress at once.




