November 18, 2014

Analyzing your Enterprise Graph to… Increase Innovation across your Organization


“The Enterprise Graph is populated by harvesting meta-data from collaboration tools, social networks, communication channels and business applications, and describes all types of people interactions.”

When you are in the Social Business business the first, second, and third question every customer asks is “how does this add value?”. IBM is no exception and as one of the largest adopters of collaboration and social technologies is always looking for us to prove (quantifiably) its value. IBM is a company that puts a premium on innovation and expects every employee, irrespective of their role, to feel impowered to innovate; from improving business processes, enhancing testing methodologies, or designing new software solutions. Therefore, when we were asked to prove the value of collaboration we chose innovation as our first proof point and looked to our Enterprise Graph to provide the answers.

When looking to demonstrate business value from interaction data you need to connect your enterprise graph with a specific business outcome and then look for correlation (or even better causation) between the interactions patterns and the business outcome. Since IBM probably generates more patents and publications than any other company, it has lots of data that track this innovation process, and connects people and ideas to patents and publications. And since it also has one of the largest and longest standing collaboration platforms on the planet, it has bucket loads of interaction data about those same people.

To “connect the dots” between our Enterprise Graph (network data) and our Patents & Publications (structured data) we did the following:

  1. We leveraged meta-data from IBM Connections which is used by thousands of IBMers on a daily basis.
  2. We applied a series of graph analytics algorithms using our Engagement Dashboard (www.ibm.com/engage) to generate individual engagement scores.
  3. We combined the meta-data and engagement scores with the patents and publications data.
  4. We de-identified the data to guarantee complete employee privacy.
  5. We used our favorite analytics tool, SPSS, to apply a series of analysis models on the resulting data.

The Results were hugely exciting and demonstrated what we’ve all intuitively known for years… we are more innovative when we combine our collective ideas, knowledge, and skills. The whole is definitely greater than the sum of its parts. To get more specific there were three main points that I would like to share.

Being Social Does Impact Business Outcome
Our analysis observed a statistically significant correlation between employee engagement and innovation results where optimally engaged employees are 120% more likely to generate measurable innovation. After employee tenure it was the single most important predictor of innovation, by a long shot, and it was only marginally behind tenure.

Optimal Social Behavior is Different for Everyone
The interaction behaviors of the most effective employees varies depending on business outcome and therefore to maximize individual impact we need to identify the optimal behaviors (usage patterns) for each desired business outcome. It’s the combination of a variety of interactions used in concert that most effectively contributes to business outcome.

Discovering and Disseminating Optimal Behaviors is the Key to Improving Overall Business Outcome
Discovery: We developed a novel methodology to discover desired business outcome & associated optimal behaviors where we trained an ensemble of multivariate analytical models to predict desired outcome, leveraged a spectrum of interactions as predictive variables, and used this to determine the behavior profile most likely to generate the best outcome.
Dissemination: We also developed and deployed across IBM an Engagement Dashboard (www.ibm.com/engage) which provides a channel through which to disseminate these optimal behaviors via our scoring engine, providing every employee private, personalized, actionable, and targeted recommendations to help them improve their impact and success.

Since the Engagement Dashboard is my personal project its not surprising that I find the last point particularly interest since it indicates that collaborative engagement as a measure (our scores) can be used as a proxy for an organizations innovation potential. If you want a more innovative organization, increase the engagement scores. It’s a slightly simplistic view that needs to be more fully validated, but its not an unreasonable hypothesis based on our data and analysis.

Follow

Get every new post delivered to your Inbox.

Join 3,056 other followers

%d bloggers like this: