Skip to content

Email is broken, and it’s time to fix it

Your alarm goes off at 5:30 AM and you reach for your smartphone, which cheerily notifies you that overnight you received 54 emails. And you only went to bed five hours ago. After a quick glance, you reply to a couple and get up to start the day. After arriving at the office you’ve suddenly receive another fifty, and so you take the first hour of the workday to sort through them and respond to those that you can. A 9 AM meeting cuts you short, but you continue to tap out a couple of replies during the boring bits. All throughout the day your laptop is dinging and popping up notifications about new and urgent request from colleagues to get them those numbers for the report, or to figure out where the best place is to get sushi. You receive several hundred emails throughout the day, and despite your best efforts, your unread count consistently hovers around 1,200 or so. Finally, quitting time rolls around and you head home, only to later take an hour or two away from your family and rest in order to try and tackle a few more in the futile attempt to get down to inbox zero.

Sound familiar? This is the doom loop experienced by many people, especially managers and executives who are suffering from information overload. No one is immune to a complete glut of messages, information, reminders, and group communications. Yet the email system we employ remains generally the same as it was over 30 years ago, when email was invented. Well, believe it or not, technology has come a long way in the last 30 years, and I believe we are finally at a point when we can begin tackling this issue head-on and improve the work lives of many people, subsequently giving them more time to spend with their family and friends.

Why doesn’t our email client tell us what needs to be done in the next five minutes? Why do we still treat inboxes like massive lists of equally-weighted unique messages? Why are we not using the resources of the cloud to apply more machine learning to the conversations we have to allow us to be smarter about them? Which behaviors are bad when it comes to communicating, and which are beneficial? Though not exactly declarative of what the future will really be like, perhaps we can glean some good ideas on communication from this video:


It is the final semester of my masters program of Information Systems at BYU, and we are enrolled in a capstone course where we have free reign to choose a project that melds all of the material we have learned over the past several years into a culminating show of knowledge. I followed a Ph.D. prep track during my degree and therefore wanted to incorporate and hone the research skills I gained in those classes. I’ve been searching for a project that would be more than just building a web app or creating a marketing plan; I truly want to start changing the world. When I saw businessmen and women who are also husbands and wives, fathers and mothers, taking unreasonable amounts of time out of their day just to try and keep the beast that is email in its bursting cage, I found a problem that finally fit the bill.

As I am still formulating the exact streams of research I want to pursue and the deliverables I want to create over the next three and a half months, I am operating at a general overview level. As such, let me just share with you some notes that I have been jotting down as I’ve explored different ideas. There is no real organization to this, and it may just be my own thoughts hastily copied down. However, I will be honing my concentration over the next few weeks and will include more detailed, knowledge-rich posts as I go. In the spirit of open knowledge, I will include all my research here for public consumption. For now, here’s what I’m thinking:


  • Email is a huge timesink for executives, and they never ever catch up
  • It represents huge inefficiencies introduced by…???
  • What do we need to do to fix it?
  • Some are beginning to claim that “email is dead” and that social will overtake it. Email will not die, but it will evolve. I want to be a part of that evolution.

Fixing it:

  • Introduce natural language processing to automatically categorize and sort messages based on their content (what project is an email referring too, what actions are required in it, is it just an announcement, etc.)
  • Apply contextual clues to the email, like the org chart and where the person and other people lie in it to assign weight
  • Remove emails out of the “line item” view, and into more a natural, project group view
  • I should be able to look at my email client and have it tell me what I can get to next, what I need to do.
  • People don’t want to spend time building filters, categorizing, or sorting. Facebook taught us this. (Can we use a social graph to add context like they do?)
  • Labels are way better than folders. What is better than labels?


  • Does Gmail’s priority inbox work for some people?
  • How do you solve the issue of a new message always popping up and distracting you from an existing task? (Well, don’t have a popup or a sound notification)
  • What do plugins like Xobni bring to the table?
  • What exactly are the inefficiencies? (I’m guessing research has covered this before)

What has been covered before?

Topics to research

  • Natural Language Processing (NLP)
  • What efforts have worked before?
  • Affordance

Call for Help

Please, if you or anyone you know has even the smallest bit of domain expertise in email, communications, affordance, NLP, or machine learning, I would love to talk with you! You can’t solve big problems in a vacuum or alone, and I will need all the help I can get. Contact me here.

Published inEmail Research


  1. I know a guy very involved in NLP. He built an article rewriter based on NLP where he used machine learning and AI to create very natural rewrites so that the article is technically ‘unique’. I could put you in contact if you are interested.

  2. You’ve got some great ideas already. One small idea I had (while not necessarily addressing the core issue you’re looking to address, I think still fits into the larger picture of interacting with the user is a smarter manner) is for email servers/clients to maintain both a state about read/unread emails and seen/unseen notifications.

    I tend to open an email, read it, and if more action on my part is necessary later on mark it as unread again. The “starred”, “important”, etc labels don’t have the same power to me as “unread” (typically lacking an overall count and feeling overall easier for me to ignore/forget about). Other times, like everyone else, I’ll see the subject or sender and not read the message until later, dismissing the notification on whatever device I happen to be using at the time (phone, tablet, desktop, or laptop). It’d be nice if I dismiss that notification, leaving the message unread, if the server stored info that I had seen an alert so my other devices would know whether or not I’d like to be notified again about that particular email (being a user-configurable option, of course).

    Perhaps I’m more of an edge case with this behavior. My original idea with this was for instant messaging, where it seems to make more sense for everyone. You’re logged into Google Talk on your computer, tablet, and phone. Google’s already fairly smart at only sending the first message of a new conversation to all devices and then limiting subsequent messages to whichever one you responded from (although something seems to have changed in Gmail recently on this point…). With a saved state, when you respond on one device, your others would get a status message from the server letting them know you’ve seen the conversation and don’t need the original notification any more.

    Just something to keep in mind if you get into the idea of making notifications about email smarter. I also love how Google decided however many years ago to keep all your chats as email threads in your Gmail account. Totally awesome and endlessly helpful.

    • Great thoughts Tim, I appreciate it. Eventually, the whole read/unread count will play a pivotal role in the overall process, so it’s awesome to hear different views on it.

      I was originally thinking of doing away with an unread count in general. It seems that, while this information may be interesting at times, it actually does little to help you accurately assess where you are in terms of processing and responding to the information. There are a lot of different ways to deal with it though, so thanks for sharing!

Leave a Reply

Your email address will not be published. Required fields are marked *