How to Personalize Email without your Data

How to Personalize Email without your Data

Knowmail is making a change in corporate communication, helping employees understand which messages to focus their energy on thus utilize time better and get the right things done and personalize email. Such priority and direction is much more complex than simply listing items per the time arrival (as is the way in email clients these days), as this ancient way means you are dealing with emails by other’s importance and not your own…needs.

This is beyond having a complicated algorithm in the background, for to maintain user’s privacy, there is a scarce amount of data that may be used, times when it is appropriate to do so and the means.

The wish list for accurate and meaningful recommendation would be email context, address book, calendar and user feedback. Per our security policy at Knowmail, the utilization of this data is extremely limited and anonymous, yet we will go over why this data is so pertinent.

Email context

As you are aware from working with your email every single day, it contains a ton of data: names, places, tasks, and actual individual details including actual attachments, which contain even more info.

This data is absolutely golden, for you can understand a wide scope, and can recreate a complete flow, including appropriate and rather mathematical recommendation. Along with historical context and ongoing correspondences and timelines, your accuracy gets better and better.

Address book

Your contacts open up a whole new world of details, which when combined with emails, can shed a lot of light on importance, prioritization, as well as upcoming requirements. It also grants a deeper view into any specific contact in terms of additional email addresses, locations, relevant links and profiles, and historical relationship.

Making such connections is crucial in understanding who is who and where they are in your email communication chain for enhanced attribution.

Calendar data

Our calendar is more than just for birthdays and special occasions, as for work it keeps us organized and within timetables, displaying meetings, deadlines, upcoming projects, follow ups and more. One could easily see how having the access to the calendar data grants so much more info for actual email personalization. Knowing specific deadlines, tasks due, upcoming meetings and more could definitely paint a great picture as what is important, as well as when it is important…as something that is a priority today is not necessarily one tomorrow.

User feedback

Imagine if you would have the possibility to tell me how important every single message is, why, and what you wish to do with it. This type of feedback wouldn’t just allow to prioritize your email, but would bring personalization to a whole new level, as it’s all about you.

Unfortunately, we don’t necessarily have the time to provide such feedback, nor do we necessarily know it, for as mentioned, things are extremely dynamic and prioritization changes from day to day.

Still, this is within the dream list of items for a based suggestion and personalizes it even further to the individual’s needs.

High integrity DNA

So we went over the Holy Grail for machine learning towards email personalization, and we are happy to note once again that Knowmail has taken a different route. Knowmail builds a model that is individual for each user. This model is created by extraction of data without any identifiable information, nor context/semantic connection, and only utilizes a set of features (parameters) decided upon prior to extraction. This means that while there is access to a plethora of data, the information utilized is limited, anonymous, and is made up of a set of items that couldn’t be reconnected for any real context nor meaning…rather an email personalization model per your habits. Yes, it makes the task much more complicated and difficult, but we are still delivering 

Meir Cohen
Meir Cohen

Meir Cohen is Knowmail's data scientist, developing and improving the algorithm and solution.

Leave a Reply

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

  1. Pingback: Should AI Have a Seat in the Judicial System? - Knowmail