Measuring Mood by Messages and Content within an Enterprise

by Richard Harbridge on March 25, 2010


Otherwise known as Sentiment Analysis.

It’s an interesting topic and openly debated as to whether or not it is reliable or effective information can be gleaned/discovered and then analyzed. In fact there is a Symposium  coming up in April (and many other events) on Sentiment Analysis.

Here are some thoughts you may/or may not find interesting on it.

Sentiment Analysis in the Enterprise:

Similar to the process of measuring mood/sentiment on twitter. Where positive and negative language is parsed and analyzed on say Trending Topics, or on specific areas of interest. This information can/does provide value to many marketing groups in the same sense a survey might have in the past. 

We can do the same thing within an enterprise. This can provide a wealth of important data to help executives, managers, and leads take appropriate steps to ensure morale and general mood around various topics remains high.

Sentiment Analysis (or a variation there-in) is Critical to a Successful Enterprise Business:

In my opinion this is crucial to a company (especially large enterprises) being successful. Whether this is done already by empathetic managers, and leaders, or whether it is aided by technology and methodologies it is still a critical part of any enterprise business.

Things that are important for an organization to understand about it’s people:

  • What motivates each person in our organization?
  • How do people feel in our organization, division, department, group, or team feel?
  • What are the goals of our organization’s people?
  • What is each person good at?
  • What is each person bad at?
  • What are our peoples opinions on specific subjects/areas of interest?

Not understanding these things within the organization results in:

  • Unmotivated People
  • Decreased Morale
  • Unproductive Organizational Culture
  • Less Profitability

Why is this more relevant now?

There is more sharing now in electronic (measurable, and minable) form then there was when everyone communicated in person. Then add in the fact we can actually store this extroadinary amount of information and already have methods and capabilities of rolling up alot of this information (social networking). People have far greater visibility in an enterprise now then they ever had before. This is growing, and becoming more and more important. For a small organization everyone most likely knows all the things I listed above about an organization or relies on a close knit group of leaders to coordinate the sharing of that information. In larger organizations this is no longer possible.

This is one of the primary reasons “People Search”, “Expertise Search”, “People Profiles” and more are a part of many Enterprise Platforms on the market today. It empowers people to connect with one another, share more effectively, and most importantly reduces the issues of having meaningful (and profitable) relationships across an enterprise.

With powerful enterprise platforms to help manage our content (such as SharePoint) while automatically tying it to who we are (our Profiles within the platform), we have more information available to us then we had before.

This also adds the most important value to our content: CONTEXT.
(Why is this relevant? How is this relevant? etc)

A CEO/Board Example:

When you think about it, at the end of the day the organizational leadership (CEO, Board, etc) doesn’t make decisions in subject areas. This is done by the leaders they appoint in those business areas. Such as the VP of finance, the CIO, or a even specific Project Managers. These people are the ones who understand the subjects they are dealing with and share the important highlights/information with the Executive or board members so that they can make educated decisions for the company as a whole.

Now imagine how effective/useful it would be to understand how the company might react/how people ‘feel’ across the organization about say: a recent restructuring effort. Take that example and break it down by geographical area, region, or sub topics within the restructuring. Then relate how this changes over time to how performance increases or profits for the organization change. Now imagine being able to take action/provide clarity, or create initiatives based on this information so that you can target areas where people are upset, or unhappy with their work and come up with a meaningful strategy for improving the business.

As they say in the customer service world:

When a customer is the most upset is when you have the best chance to make them a happy customer for life.

It works similarly for employees.

With the powerful Business Intelligence capabilities we have available to us, and the new ways we can digest massive amounts of information is this unrealistic? I don’t think so.

In my example above I target the highest levels, but you could probably imagine uses for this sort of data in many areas of the business.

The words: Talent Management 2.0, Succession Planning, Human Resources, Employee Retention, etc are all things that come to mind.)

Let’s take it one step further…

In SharePoint 2010 one of the interesting new features is the ability to ‘mine’ your Outlook Sent Items for suggestions of what your interests might be. (Don’t worry this feature can be applied in all sorts of different ways, to find out more about Social in SharePoint 2010 take a look at my recent presentation.)

When I talk about gathering sentiment analysis data I don’t just mean from your Enterprise Platform (documents, usage statistics, discussions, notes, tags, blogs, etc) but also gathering it from Exchange as an example. In the same way we might suggest interests, we could also ‘suggest’ (in a different sense) an individuals feelings relating to a specific topic.

Extend this potentially to IM conversations, and maybe even phone analysis.

What are the challenges?

  1. We have to have people comfortable with “Sharing” how they feel in an enterprise.
    One of the biggest challenges of Sentiment Analysis is that most of the time people are neutral with how they feel. I mean it is ‘professional’ and still a business. So it’s typically more rare for an employee to show emotional context in professional communication. With how blended personal and professional lives and communication are becoming this might start happening faster than you think.
  2. We need better Sentiment Analysis Algorithms and more accurate results.
    This is still a new and very young field of analysis. With anything new it’s difficult to tell how effective the results and more importantly application of Sentiment Analysis can be. Most show an accuracy level of 70% or so in measuring mood. This isn’t a reliable enough number yet in my opinion.
  3. Greater maturity and understanding of how to make effective decisions with the data is necessary.
    It’s difficult enough to make decisions with the wealth of data we have available. Trying to take a concept like this and make effective decisions with it is even harder and more complex.

It’s an interesting blossoming area of intelligence and I know I will be interested in learning more as it develops. For anyone curious Microsoft does not have a partner who specializes (or from what I can find) even provides this sort of analysis. Opportunity?

What do you think?
Richard Harbridge

VN:F [1.9.22_1171]
Rating: +3 (from 3 votes)
Print Friendly, PDF & Email

Previous post:

Next post: