How to triage your inboxes
Hi everyone,
Today I thought I’d tell you about a useful thing I’ve ended up explaining to a bunch of people this week. It’s a missing insight for a lot of people’s workflows, so fits nicely into the How to do everything theme.
What’s the problem?
Back in Decoupling I talked about the problem of inbox management: You’ve got thousands of emails, or books piling up to be read, or things to declutter, or articles to read, or… You get the idea.
Because you have so many items, the whole thing feels overwhelming, so I suggested that one way to approach this problem is to stop equating the whole thing with the individual parts and tackle it one at a time. If you can’t answer any emails because it feels just as hard to answer one as it does to answer every email, just answer one email.
This is a useful psychological strategy in some ways, but it has a couple of key issues. One of them is that inboxes, by their nature, tend to build up over time. There’s a queuing theory problem here: Either you, on average, process your inbox faster than things enter it (in which case it is typically empty) or you process it slower than things enter it, in which case it grows over time.
Processing your inbox one item at a time genuinely isn’t guaranteed to keep on top of things, and the result is that just doing it at a rate that feels comfortable might genuinely not be enough. As well as comfortable strategies you need strategies that actually keep up with the inbox. As well as practically useful, this is emotionally important - otherwise you will feel overwhelmed by your inbox, because you are, and the whole thing becomes aversive and the problem becomes worse.
So, how do you stay on top of an inbox?
Triage Models
On average, how long would you say you spend reading a book on average?
Trick question, the answer is probably in the region of one second.
If you believe the estimates that I got from about 30 seconds of googling, there are probably somewhere in the region of 130 million books that have ever been published (this is probably an underestimate as these figures are from a few years ago). If you spend about two hours a day reading for sixty years, you get to about 160 million seconds spent reading over the course of your life. So you’ve probably spent a bit more than one second per book under that estimates, or less than one second per book.
Of course, some books you spend considerably more time on than this, but even there there’s a similar process of triage. How much time do you spend on reading a typical book you’ve actually encountered? My guess is that it’s not as high as a minute, especially if you’ve spent any amount of time browsing book stores. We encounter a lot of books.
Generally there is a pipeline that goes something like:
Never encountered this book (0 seconds)
Judged book by cover, never opened (1 second)
Flicked through book briefly (30 seconds)
Read enough of book to decide whether it’s worth reading more (5 minutes)
Read some of the book, didn’t finish (an hour or two)
Read the whole book (typically somewhere up to ten hours)
Reread the book multiple times or extensively studied it (anywhere between tens and hundreds of hours)
With a huge drop off rate in the numbers of books in each category. Most books you never see, most books you see you never open, most books you open you never read in depth, most books you start reading you never finish1, and most books you finish are not worth repeat study.
I tend to refer to this approach to things as a triage model - you are constantly using your time to decide how much more of your time you should be using. I think about a lot of processes in these terms, and this attitude (though I didn’t use the term there) came up before in Mathematical Cranks, Information Overload, and Predicting the Pandemic as a way to deal with the problem of there being far too much information to verify all of it.
Stage-first triage models
So obviously my advice is to apply the above model to triaging your inbox - spend time on each item to decide whether it needs further attention - but you were probably going to do that anyway, whether explicitly or implicitly.
But one thing you might not have thought to do (certainly it is a thing that it took me a while to figure out and most people I talk to about this don’t do it) is to break down tasks by triage stage.
If you take an item first model you will tend to progress each item through the stages until it is done. You spend attention on the item until it is done - first a couple of seconds to decide whether to archive it, then 30 seconds to decide whether it really needs a response, then five minutes to respond to this.
I’d like to convince you that this is often the wrong way. It’s fine if you don’t have that many items to deal with, but as soon as an inbox starts to look overwhelming this will contribute to the problem.
The problem is that basically context switching breaks flow, because you’re constantly having to switch between different approaches and attitudes to things. If you are doing many things of the same type, you can very easily get into a rhythm. If you are constantly having to change task types, you can’t, and the whole thing both takes longer and feels more laborious.
Instead what you should do is split up your inbox work by triage stage. Focus on everything that is in one triage stage and work solely on the question of whether to move it into the next stage. This is a stage-first triage model instead of an item-first triage model. You organise your model around stages, not individual items.
(In real life triage, generally you take advantage of this by having multiple people working on the problem, with individual people dedicated to each stage, but I’m assuming there’s only one of you).
Initial triage
If you have a “raw” inbox - one where other people can submit items to it (email, physical mail, an RSS feed, twitter notifications) - the better way to process it is not to try to answer every item one at a time, but instead to allot an amount of time in which you spend at most a couple seconds per item making one of two decisions:
This does not need action and can be disposed of (mark read, archive it, throw it in the bin, etc).
This might need action and should be put in the next “inbox” (e.g. open in a browser tab, star it, a literal physical tray, etc)
If you’re not sure, it might need action, therefore it goes in the second category. There is no reason to spend more than a couple seconds per item, because once you have spent a couple of seconds on an item you have implicitly made the decision that it is worth spending more time on. Move it to the next inbox and move on.
There is one exception I sometimes apply to this timing rule: If your reaction to an item is something along the lines of “Not only does this not need action, nothing like it could ever possibly need action” and there is an obvious and relatively quick thing to do to unsubscribe from such items, it’s worth taking the time then and there to do it. Update your email preferences on a site, unsubscribe from a feed, etc. It breaks the flow, but it’s worth doing while it’s in front of you and hopefully shouldn’t happen that often (and will happen less over time).
This can be combined well with the decoupling approach in that your decoupled task can be to spend a fixed amount of time triaging it, but honestly you will probably be surprised at how little time this step needs. If you have 1000 items and can process at a rate of a second per item (which in many cases you can), you’ll be done in 20 minutes.
Some worked examples
After the first stage what you need to do depends a lot more on the specifics, and honestly whether I’m good at it or not depends a lot on which area it is2, so rather than providing general principles, here’s a worked example.
The area where I’m most reliably good at this is my RSS feed reader, where I liberally subscribe to things on the internet that I find interesting, and then when I fail to keep on top of it for a few weeks suddenly find I have 500 unread items to deal with.
Here is my specific process for dealing with that:
Open my feed reader in a dedicated browser tab and go to the top of the “unread” folder, with no differentiation among different sources of items (I do have some categories, but by the time I have this problem the categories are irrelevant and I just want to get down to inbox zero).
Starting at the top I use the keyboard shortcuts to navigate. “j” takes me to the next item, which I can see the title of and usually read at least a paragraph or two of. I look at it, do a gut check of “Does that sound interesting?” and if so press “v” to open it in a new tab, then immediately tab back to the feed reader.
Once I have gone through all items like this, I close the feed reader. I now have a browser window open with a set of tabs that I intend to maybe read.
I go through each tab and spend about 30 seconds on it at most, reading through the beginning and deciding whether or not I want to read more. If I don’t, I close the tab. If I do, I move on to the next tab.
Once I have been through each tab at least once, I now have a window that probably has only a handful of open articles in it. Fewer than 10 almost certainly.
I now go through each of these in turn and read them, closing them when I finish or get bored. Or, realistically, I get distracted and leave it for later.
Not coincidentally, I also have a triage model for browser windows and tabs. It goes as follows:
Go through each open tab and close it if it I can’t think of any possible reason why I still need it open.
Consolidate tabs into windows that roughly group them by purpose or intent (e.g. stuff I might read later)
Spend some time reading through things closing them as I go.
If something is too much work but I still want to read it later, I either leave it open and move on or physically print it out for later consumption.
(Yes, I also have a triaged model for physical print outs. Yes, it involves a lot of things going into the recycling unread)
Internalising triage models
As well as being practically useful, these staged triage models are really helpful for reframing our relationship with work, by giving us a better sense for how long things actually take.
I think what often happens is that we end up anchoring our internal time estimates wrong. Consider:
It takes me five to ten minutes to reply to an email.
I have thousands of unread emails.
Therefore it will take me hundreds of hours to reply to all my emails! Tragedy and woe!
But the reality is that it doesn’t take five minutes to reply to a typical email, because a typical email doesn’t require replying to at all, you just don’t notice the relatively trivial amount of time you spend on the things that take a couple of seconds. A typical email that has made it to the final triage stage of genuinely for reals needing action takes five to ten minutes, but the typical email never makes it to that stage.
The problem is that our notion of “typical” is weighted by how long things take, so our notion of how long each item takes naturally skews towards thinking there is a lot more to do than there actually is.
By dedicating time to triage stages individually, we re-anchor to a different notion of typical. We treat the triage as its own atomic activity with time dedicated to it, and we can get a pretty good grasp of that as its own thing, with a more accurate intuitive sense of how long it takes.
Adopting this model can’t guarantee that you will no longer feel overwhelmed by your inbox. This would be an irresponsible thing to try to guarantee because it might just genuinely be the case that your inbox is intrinsically overwhelming. But by adopting this model you should be able to start to feel an accurate sense of how overwhelming your inbox actually is, and develop better habits for dealing with it.
Coaching Auction
Every week I auction off a coaching session based around the Wednesday free newsletter issue. We spend an hour talking about something loosely based on the theme of the issue and discussing relevant ways you can use it to improve your life. You can bid for this week’s coaching session here. I must admit I’m not entirely sure what that will look like with this week’s theme, so feel free to go quite broad - if there’s something you want to get better at, we can work on ideas for how to do that even if it’s not literally the method talked about here. I guess I’m also happy to coach you on Slay the Spire if you really think that’s a good use of your money.
The auction uses a mechanism called a Vickrey auction, where you submit a sealed bid, and the winning bid pays the price of the second highest bid (so you never pay more than you bid, but will always pay less unless there is a tie, in which case you pay the amount that you bid). Typical winning bids are in the £50-£100 region. Last week’s Record yourself to coach yourself, the winning bid was £50, paying £50.
In addition, although I’m currently reluctant to take on new recurring clients, I do offer one off troubleshooting sessions, currently at £95 for an hour session, so if there is some problem on your mind that you think I could help you with (anything from feelings to software development), feel free to drop me an email at david@drmaciver.com to enquire.
Credits
Image credit for the preview of this week’s post goes to Brendan C for his picture “Anxious Pile”.
This is not true of a lot of people and if it’s not true of you, you should fix that. There is no obligation to finish a book, and your reading will improve if you do not behave like there is.
Despite my use of “inbox” here I promise I am not good at staying on top of my email.