Despite the rise of messaging apps such as WhatsApp and Slack, email remains the primary method of communication in 2020. The rumoured demise of email has so far not materialised, yet the challenges remain. Email overload is a real problem and whilst the issues are well documented, email remains a vital part of everyday life.
What makes email so useful?
Email remains popular because of its dual-use as an archiving tool. Users are able to file messages and retrieve them at a later date – often in real time. Secondary is the relative simplicity and low cost of email. Using email requires very little, if any, training and is usually free.
The problem of email overload
Finding a specific email can be time consuming. We exchange huge volumes of emails on a daily basis making searching them difficult. Manual filing and sorting of emails can help but over this can create an overwhelming burden for the user.
Classification of emails as a solution
Teaching machines to automatically classify and summarise emails is one solution. This is achieved by looking at the highest frequency of words in the body of an email and using these to assist the user. In fact, users often complain that the majority of email received is junk mail or spam. Therefore the opportunity to automate the process and weed out the emails which have little value, is huge.
Detecting value
As a simple starting point, the use of machine learning to detect sentiment or commitments can be hugely productive. Imagine if you could focus solely on those tasks you had promised to do. Or, if you could immediately identify the most disgruntled customers first so that you can tackle these as a priority.
For example, phrases such as “I am pleased,”, “I am happy” may infer a more positive customer experience than “I am disappointed” or “I am unhappy”. The ability to help designate emails in order of importance could not only improve customer service but also help to train individuals.
Dealing with the clutter
Artifical intelligence can also be used to de-clutter your inbox. It can make achieving inbox zero a reality rather than just a theoretical possibility. There is a misconception that email clutter is the same as junk email. But that is incorrect. An email sent to multiple users about a support issue is of value to the team member dealing with support but is of less importance to a recipient not involved on the matter.
By learning from your own inbox habits, this technology can revolutionise the way we handle and deal with email on a daily basis so that we focus on the emails that really matter.
The challenges of privacy
One of the most common challenges when reviewing emails is the inclusion of personal data. From the obvious: an email address or name; to the less obvious: the actual content in the body of the email. The subject of privacy is a common concern for users who do not trust the ability of a machine to decide which information is confidential or sensitive.
But research has shown that a high percentage of emails exchanged on business email addresses with third parties are in fact not confidential. In fact, it is usually internal emails that are likely to contain sensitive information. Yet most employees are already aware of and bound by email use policies which means that will normally exchange personal emails using a different email address or other methods of communication e.g. SMS, WhatsApp.
Conclusion
It is clear to see how machine learning can be potentially useful to solve the problem of email overload. A trusted system that learns and understands the behaviour of the individual user is invaluable in helping businesses improve productivity. The starting point however is to capture email data from across organisations and use that data to better understand user habits.