What Email Reveals About Your Organization: Reading Summary

Summary:

The article my group read this week was entitled ‘What Email Reveals About Your Organization’ by Peter Gloor. As you may have guessed, the primary purpose of it was to show the kinds of insights managers can learn by studying how their employees use email and some of the best practices associated with it. Peter has worked as a researcher for fifteen years studying various organizations through their social networks. His goal is to develop enterprise software that will allow organizations to track informal knowledge flows in the same way that existing software is able to track things like financial information or business process flows.

Since his studies often dealing with mining large email archives, he first pointed out the steps they take as researchers to quell privacy concerns:

  1. Commitment to doing anonymized analysis
    • they aggregate results by team or business unit, only individuals can see their own communication patterns
  2. Restrict most content of our content analysis to email header information, which includes the sender, receiver, subject line, and timestamp
    • Machine learning software analyzes message content for sentiment and emotionality
  3. Commitment to transparent communication with research partners and management
id20151230mgmsg-a

Visualization of an Organization’s Social Network

The article then went on to reveal the several indicators of effective collaboration the researchers discovered as a result of their analysis:

Type of Leadership

Instead of everyone in a group acting as a leader, for creative work it is more effective to have strong leaders that take charge of a group. Even more effective are groups of leaders that rotate depending on the specific project involved. Conversely, when dealing with work where reliability is more important than creativity, steady leadership is more effective than rotating.

Participation Level

There is a difference between information consumers and producers. Teams whose core members contribute a similar number of e-mails are more creative than when just a few members produce information. However, sometimes customers prefer to communicate with just a few members of a team rather than many.

Response Time

The speed of response and the number of ‘reminders’ required to get a response tend to be good indicators of employee satisfaction and mutual respect. Happier customers tend to answer emails faster.

Language Tone

When analyzing message sentiment and emotionality, they found the more positive language a salesperson used with a customer, the less happy the customer. Employees likely to leave their jobs became less emotional in their emails, contributed less leadership in months leading up to departure.

Shared Context

High functioning teams define their own language. The more complex language a salesperson used, the more dissatisfied the customer. The more successful someone was in introducing new words to a team, the more influential that person was in the context of the group.

gloor-email-organization-1200-300x300

Finally, the article outlined a four-step process to improve the performance of organizations:

  1. Define social network metrics and communication patterns
  2. Compare structural attributes with business success
    • Correlate indicators with success and failure metrics specific to the organization
  3. Mirror behavior back to individuals and teams
    • Help improve employees, show them how they can improve
  4. Devise a plan to optimize communication for greater success
    • Help managers figure out how to change employee communication behaviors to lead to better success in the future

 

Discussion and Analysis:

Personally, I didn’t really think the article was that groundbreaking. To me it seemed that the five ‘indicators of collaboration’ were all essentially common sense. Additionally, the article qualified each indicator to work in specific situations but several times said the opposite would work in a different situation. For example, in terms of participation level it said creativity is higher when all members of a team contribute equally but in certain customer service situations the opposite was true.

Going off of this, a point that came up during our group discussion was that no work situation could 100% fall into either a ‘creative’ or ‘reliability’ label. Every job would include aspects of both creativity and reliability, meaning the article’s findings are sort of moot.

Another point that came up in our group discussion was the study was leaving out the most valuable aspect of the e-mails. While we understood the privacy concerns associated with analyzing the actual content of the messages, at the same time we felt this content could reveal a lot more about the effectiveness of an organization’s social network and even an individual employee’s productivity. Going beyond just analyzing the messages for things like emotionality, we felt the study could have revealed a lot more had they not left out the message content.

7 comments

  1. I think you did a great job of summarizing our reading along with our discussion in class. I agree that I didn’t find the article very groundbreaking, but I do think the author did a great job of separating the indicators of effective collaboration into a clear and concise list. I also think that the final four step process for improvement is a beneficial read for any company in today’s world.

  2. Nice job summarizing your group’s article. I was in a group that didn’t read this article, but I decided to because it sounded so interesting after the class discussion. You did a great job hitting all the key points and I liked hearing your own analysis on the issues. I think of my own workplace and past jobs, and I liked reading about the types of leadership for various jobs (creative versus reliable). I definitely agree and I’ve worked at jobs where creative projects had weak leaders, which resulted in not-so-stellar results. Enjoyed reading this!!

  3. I completely agree, Justin. My group also read this article and we had many of the same points. I see why companies would want to analyze email, for it’s an activity that takes up a large portion of its employees time. However, I agree in that the parts of the email that were being looked at didn’t seem to carry groundbreaking information. I believe this study had the right intention but could be executed more effectively in the future when there’s better data and insights around qualitative content.

  4. The two things I meant to mention (but didn’t) in class was 1) that I suspect that they are developing tools that allow managers to do these things in real time and 2) they do have technology that can analyze content while protecting privacy. Basically, the computer gets to see what you’re talking about but it doesn’t tell your manager.

  5. Nice breakdown! This article seems to nicely contrast the article my group read about emailing being a waste of time. On the one hand there are more efficient ways to communicate and accomplish things like scheduling, but as a sort of catch-all platform, email can provide at lot of information about a company. I think what Professor Kane said about analyzing the actual content makes a huge difference. I still don’t know if I’d want to spend most of my time at work on my email when I could be more productive elsewhere and it’s also being analyzed.

  6. i am not a huge fan of email. Over the summer i interned with NYL and they used a chat application that was much more useful than email. Mass communications were still done over email. but all other correspondence was done over a chat feature that worked similar to imessage. realistically, if email is ever replaced i think it will be by something drastically different from email as we know it

  7. ajsalcetti · ·

    Nice summary. I do agree that the five indicators seem pretty common sense, but sometimes those are the easiest ways to filter information for end users and make sense of everything. As they say, the devil is in the details and while we know leadership, participation, response time, etc are indicators, it is how those are being used and not just if they are being used. The machine learning we have discussed in class then analyzes and makes sense of everything for management. I know personally I rarely put any effort into my headers, as long as it makes sense enough to the counter-party. Then I give the “answer” in the response, which would have the juicy details of content, sentiment, emotion, etc. So in its current form, I agree that a computer analyzing the send, subject line, timestamp, etc isn’t going to get a lot (especially from me), but as Prof Kane mentions in the comments area, there are computers that can anonymously analyze content and that will provide much better results.

%d bloggers like this: