Technological Solutions to Email Overload

Thanks to the years of research that have already gone into email overload, there have been many proposed and experimented solutions that attack the problem in one way or another, as well as many companies springing up today. I wanted to try and give a brief treatment of what has already been covered so that we know where to turn our efforts in the future.


Past Research

Determining Message Value, Priority, and Criticality

Eric Horvitz has led the way in much of the research regarding decision-making as it relates to email. One of the most intriguing solutions proposed and tested by Horvitz and his fellow Microsoft researchers was the Priorities system they developed to determine the criticality of a message upon receipt using Bayesian networks, and to take certain actions based on that determination. Their research not only covers an important factor in determining the overall value of a message, but also helps to reduce the costs related to over-alerting an individual with email updates and the task-switching costs that are incurred. Likewise, they also discussed the costs associated with delaying notification of a message.

The Priorities system they designed was implemented for real-world testing at Microsoft. I am not sure what has been done with the system or the work since then, but it was stated in the article that it seemed to be a success in their testing.

See: Horvitz, E., Jacobs, A., & Hovel, D. (1999). Attention-sensitive alerting. Proceedings of UAI  ’99, Conference on Uncertainty and Artificial Intelligence (Vol. 98025). Stockholm, Sweden. Retrieved from http://dl.acm.org/citation.cfm?id=2073831

Even earlier work was done in 1993 by Douglas Terry, a researcher at Xerox PARC, that used hand-crafted “appraisers” to determine the priority of a message. The user would create different appraisers (essentially rules) that stated, for example, to give a message a score of 95/100 if it was sent by the user’s boss, or a 15 if it seemed to be a mass message from HR. The email client would then sort by binned ranges of priorities. This is certainly an early and rather static treatment of priorities, but could potentially be expanded now using machine learning and more advanced natural language processing to be more scalable and powerful.

See: Terry, D. B. (1993). A Tour Through Tapestry. COCS  ’93 Proceedings of the conference on Organizational computing systems (pp. 21-30). ACM. Retrieved from http://dl.acm.org/citation.cfm?id=168558

Different Paradigms and UI Presentations

One of the more complete treatments of a new messaging paradigm was the TaskMaster system created by Bellotti, Ducheneaut, and Howard. The TaskMaster system was “intended to break away from the messaging-system metaphor (featuring inbox, outbox, and drafts folders) and to present a strongly-focused task-centric view of e-mail.” As such, emails are viewed more as tasks, and the client relied heavily on threading to organize the tasks (they gave this a new name: “thrasks”). They also elevated document attachments and links to be “first-class citizens,” previewed within the same window as the message itself. See the screenshot below for an explanation of the UI.

Screenshot of the TaskMaster UI

This new email client was tested by real users, and the reaction was overall quite positive. However, they found that most users didn’t stay with the system long term because the client was not as full-featured as the client the users were already familiar with (Microsoft Outlook), and they needed the additional functionality. Using both at the same time didn’t make sense.

This brings up a very important point that is evident over and over again when new software solutions are tested: users often like the new solution, but cannot stick with it long term because they need the more full feature set of existing clients. A lesson we can learn from this is that if you are going to present a new software solution, it either has to be built into existing clients or act as a full blown replacement for the clients with feature parity.

See: Bellotti, V., Ducheneaut, N., & Howard, M. (2005). Quality versus quantity: E-mail-centric task management and its relation with overload. Human-Computer Interaction, 20, 89-138. Retrieved from http://portal.acm.org/citation.cfm?id=1466574

Another radical departure from the typical listed inbox was the TeleNotes system designed by Whittaker and colleagues. It took an approach more akin to the physical world, allowing emails, tasks, and documents to be scattered around a desktop in a visual way, similar to how employees will organize different piles of papers and to-dos on their desks. This appealed to users, and for large projects and tasks it was very useful. However, users didn’t want to use the TeleNotes system for all of their emails because many of the smaller, one-off emails didn’t fit well into the paradigm. Also, we again, saw the issue with users abandoning the new system because the prototype wasn’t fully featured, and the reliance on the old client.

See: Whittaker, S. (2005). Supporting collaborative task management in e-mail. Human-Computer Interaction, 20(1), 49–88. L. Erlbaum Associates Inc. Retrieved from http://portal.acm.org/citation.cfm?id=1466573

Organization: Automated and User-Directed

One thing we learn from extensive research is that filing into folders, even though a relatively common practice, is a rather unsuccessful technique for reducing email overload. We could potentially automate this, but that doesn’t help much either. Steve Whittaker summarized this rather well in his 2005 paper:

“Filing is a major problem for e-mail users: Categorization is a cognitively difficult task (Lansdale, 1988; Malone, 1983; Whittaker & Sidner, 1996), made yet more difficult because folder categories change as the user’s work focus shifts (Kidd, 1994; Whittaker & Hirschberg, 2001; Whittaker & Sidner, 1996).”

“Several agent-based systems have been designed to provide assisted filing (Boone, 1998; Cohen, 1996; Mock, 2001; Segal & Kephart, 1999; Takkinnen & Shahmehri, 1998). These systems use machine learning techniques to automatically elicit the defining characteristics of existing folders, based on message header properties and content. … More seriously, assisted filing does not get to the heart of the problem. Assisted filing classifies inbox messages into existing folder categories, whereas a major user filing problem is with defining new folders (Whittaker & Sidner, 1996).”

Even if the categorization is correct 85% of the time as was found by some of the implementations, there are two problems: first, users don’t trust an automated system, and second, filing is not a very good practice to begin with because it is cognitively demanding, and so doesn’t lessen our email overload.

See: Whittaker, S. (2005). Supporting collaborative task management in e-mail. Human-Computer Interaction, 20(1), 49–88. L. Erlbaum Associates Inc. Retrieved from http://portal.acm.org/citation.cfm?id=1466573

It is very important to note that, as email volume increases, having a good search function in an email client is essential. I would almost propose that our email inboxes are becoming “mini-Internets” in a sense, where we get on and search for bits of data that we know is hidden away somewhere in those troves of email, much as we open our browsers and use search engines to query the entire Web. While examining organization in inboxes, Fisher, Brush, Gleave, and Smith discussed the importance of search to many users with high volumes of email.

See: Fisher, D., Brush, A., Gleave, E., & Smith, M. A. (2006). Revisiting Whittaker & Sidner’s “Email Overload” ten years later. CSCW  ’06 Computer Supported Cooperative Work (pp. 309-312). Banff, Alberta, Canada. Retrieved from http://dl.acm.org/citation.cfm?id=1180922

Classifying broad email types is something that is becoming rather feasible thanks to powerful algorithms and data mining techniques. A more recent project developed a system to automatically classify customer complaint emails from other general emails, with rather impressive results. The setting for the research was a call center that handled customer emails, and so in this case, triaging the emails could prove to save time and money.

See: Coussement, K., & Vandenpoel, D. (2008). Improving customer complaint management by automatic email classification using linguistic style features as predictors. Decision Support Systems, 44(4), 870-882. doi:10.1016/j.dss.2007.10.010


Current Solutions and Companies

AwayFind

AwayFindAttempting to help with attention management instead of inbox management, AwayFind is trying to target the top 2-3% of messages that you need to know about immediately, and will notify you through other means of communication (text, phone call, etc.) of the arrival of such messages. It appears to do this through a combination of examining most-contacted recipients and by allowing users to specify certain rules to trigger the alerts. Their tagline is “stop checking your email. The founder and CEO of the company, Jared Goralnick, is also the EVP of the Information Overload Resource Group, a collection of practitioners and academics attempting to solve the problem of information overload, and so you can bet the company is probably on the right track.

OrchestratorMail

OrchestratorMailThis solution is particularly interesting to me as I think it goes after a very specific but very beneficial pain point—establishing a common ontology for email communications and attaching deadlines and reminders to existing conversations. OrchestratorMail builds on top of existing email clients, so it doesn’t require users to adopt entirely new software, and it doesn’t try to automate something that doesn’t need to be done, such as filing. According to the company, most email communications can be broken down into questions, discussions, notes, information, requests, and proposals. Users mark emails as such, and then when replying, instead of communicating their response only through the text reply, the user will click a button (like “approved” or “more information”) to indicate how the conversation has gone. Deadlines and reminders can also be attached to conversations. All of this extra information helps to speed up the conversation process, and remove a lot of the manual work involved. I think this particular application, while small in focus, has a lot of potential.

Asana

AsanaTo understand Asana, you can think Basecamp with super-tight email integration (especially with Gmail/Google Apps). It plays off the idea that email inboxes have really morphed into task management centers (which is true, and can be backed up by large amounts of research). Emails can be funneled through the app, and when everyone in an organization is on the system, they can assign tasks to each other, keep up to date, etc.

EmailTray

EmailTrayThis company, EmailTray (formerly SenderOk) has created a new email client with a focus on intelligently ranking emails based on contact patterns, as well as combining information from social networks to enrich your contact’s details. I haven’t tried it, but hopefully the new client overcomes the problems found earlier in the research section about users abandoning new software because it was feature rich enough.


There are of course many other solutions and proposals, both in research and in industry, that aim to help ease the email overload burden, but these are few of the large ideas that I have come across. Based on what I have seen so far, I am inclined to say that the big solutions to email overload lie with behavioral changes and improvements, rather than with technological advances. However, technology is the perfect tool to reinforce good behavior. This is all a topic within itself however, so more will be forthcoming on that.

Please feel free to mention other ideas and companies that you may know of as well that are tackling this immense issue from the technology side. We need all the help we can get!

Share and Enjoy

  • Twitter
  • Google Plus
  • LinkedIn
  • Email
  • RSS
  • HackerNews
  • Instapaper
  • StumbleUpon
  • Facebook

7 thoughts on “Technological Solutions to Email Overload

  1. Thanks for accumulating this list, Joshua. I’m only now seeing that you referenced AwayFind here, thanks! There’s a LOT more to add to this, we can chat one of these days about that, perhaps : ). Have a great day!

    Reply
    1. Post author

      Of course! I’ve been talking up AwayFind with a lot of people recently, specifically because it’s great at addressing the question “What happens if an important email between email checking scheduled times?” There are of course a ton of other uses for it too, but that one fits perfectly. Looking forward to a chat with you.

      Reply
  2. Josh – I love this blog. We’re developing a solution to email overload that centers around the visual component – mainly that it’s easier and more intuitive for users to recognize brand and personal emails by icon rather by lines of test. Check us out at http://www.philterit.com.

    Reply
    1. Post author

      I really like what you’re doing there with the visual representations. The next step would be to move beyond brands and start being able to pile messages together based on project, team, topic, etc. This has been tested in the past in some research, and has worked quite well. Any plans along those lines? And plans to make it an extension/plugin instead of a separate service?

      Reply
  3. Darryl

    This is a great summary and in line with my feeling about email. I, too, view my Gmail as a repository of gold that I can mine through search. I don’t use folders for the reason stated: too much mental energy. Search beats sort everytime, anyway. Finally, I’ve been testing Asana and would recommend it to someone with a lot of ongoing tasks and/or as a kind of task-assignment system to several employees.

    Reply
  4. Pingback:

  5. Josh, very interesting post — I thought your readers might find http://www.activeinboxhq.com interesting as their approach has been similar to some of what you’ve talked about in the post. I’ve been using this for years, and agree that with email it’s more about us being broken than the technology we have available and at the same time the tech could improve too. What I love about ActiveInbox is that it was originally inspired by the Getting Things Done model by David Allen — but it’s still very flexible with the ability to create action lists, project lists, etc and mark emails with these labels to have them appear within the review system…. as long as YOU work the system, the system can really work. They are still actively developing and always polling to see how things are working for users … I like that part too, continual improvement! Often solving problems, that I wasn’t even aware I had yet too.

    Reply

Leave a Reply

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

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>