Last night I read the paper To Reply Or Not to Reply: Predicting action on an email message, and took some notes on interesting information as I went through it. 

The paper covers a survey done to examine what are the characteristics of an email message that would indicate a users action on the message. If examining a broad context of users the survey data is a little weak, for example the participants in the survey (121 people) are all university faculty, staff, or students; 76% of whom were male.  So the data is not completely representative of all email users, however it does give us some fantastic insight on a range of users who use email for work.

Read it for yourself, but here are some highlights.

4 Distinct Purposes of Email

These were identified from lots of previous literature they reviewed about email in an organizational context.

1. Task Delegation / Project Management / Reminders
   action requests, status updates, and meeting and deadline reminders
   often left in the inbox until acted on

2. Information Exchange, Storage, and Retrieval
   information requests, and information responses
   responses often saved for later retrieval, either archived or left in the Inbox

3. Scheduling and Planning
   meeting requests, and responses to meeting requests
   email has mostly replaced phone calls and memos to become the scheduling supporter
   often quickly responded to

4. Informal Communication
   instantaneous delivery, and rapid response to email communication similar to instant and social text messages
   likely to receive a quick response

6 Email Content Types

Broken down from the 4 purposes of email in an organizational context.

1. action requests
2. status updates
3. reminders
4. information requests and responses
5. scheduling requests and responses
6. social content

Method: 2 Part Survey

Statistical Data Collection

Asked the number of messages sent, received, sitting in Inbox, and other habits

Questionaire

Asked for detailed ratings of 5 new non-spam messages in their email Inbox

Also indications as to the nature of the content, message importance, sender characteristics, and action taken on the message (replied, plan to reply later, do not plan to reply), or what they did with the message (delete, file, or leave in Inbox)

Hypothesis: Sender and Content

Users evaluate message importance based on characteristics of the message sender and content.

Results: Sender Characteristics

Sender and content are important in generating the users perception of the importance of a message.  However it is the senders characteristics (co-worker, friend, boss, etc.) that mostly determine the probability of a response to any message.

The majority of individuals (75%) had less than 1000 messages in their Inbox, however Inbox size combined with the high number of folders reported seemed to suggest people are filing messages into folders.

Interesting Stats

Out of the total messages surveyed (581)

(graph courtesy of paper)

• 36% of messages contained important content (attachment, link, or phone number)
• 34% of messages contained a request for action
• 14% of messages contained scheduling content

(image courtesy of paper)

• 64% did not require a reply (30% delete these messages)
• 23% required an immediate reply
• 13% required a reply they postponed (of these 79% were left in the Inbox)

Rate of Return Stats

• Messages sent to a list of individuals were 18% less likely to receive a reply
• Messages with only 1 recipient were 20% more likely to receive a response
• Having a work relationship with the sender lowered the probability of response by 9%
• Frequency of communication did not play a direct role in the probability of response