The Science of Innovation: A Quick Q&A with Lisa Purvis on Initiatives and Hypotheses

Filed Under Mindjet

Lights of Numbers
Arwen Heredia

October 28, 2013

Repeatable innovation isn’t something that just happens on its own — establishing a process to bring the best ideas to market depends on research, data, and the examination of various strategies. In this series, we discuss the science of innovation with our experienced internal Data Scientists, who will walk us through their innovation hypotheses, findings, and unexpected discoveries.

In this Q&A with Mindjet Data Scientist Lisa Purvis, we chatted about a few key data initiatives her team is putting in place for Mindjet. Below are the hypotheses her team formed about various aspects of innovation from looking at global data across customers, and the results and implications that are driving our products and innovation forward.

What is one of the foundational hypotheses that you’ve formed about innovation in organizations in your work?

One of the first things we hypothesized is that throughout an organization there are natural born innovators – those who just naturally stand out in their effective creation of successful ideas. We also hypothesized that pinpointing those people based on their innovation interactions would be valuable to organizations, in order to surface people that otherwise can easily get lost in the shuffle. Like a Myers-Briggs indicator, but for innovation ‘personalities’ or strengths — to enable people and organizations to know and leverage their innovation strengths.

How did you go about testing and validating this hypothesis?

First we began to explore how we might be able to uncover these innovation strengths simply based on the data in an innovation network of people, ideas, and the interactions happening between them. We created a model, based on pattern recognition, that examines the patterns of interactions happening in the network, and classifies people as innovators if a certain pattern of interactions around their ideas takes place consistently. We then applied this model across our networks, to understand the typical make-up in terms of innovation strengths, and validated the innovators surfaced with one of our key customers.

What were the results that you found in doing so?

We found that indeed, the model enables us to classify certain strengths based on ones innovation activities and behaviors. In many cases, it even surfaced some individual contributors that were strong innovators that were otherwise under the radar. We found that the innovation strength in our networks is typically only possessed by on average 10% of the user population — it is truly a unique skill.

We also looked for a second type of skill using this model:  a “discerner” — someone who is good at spotting the successful ideas of others. We found that this skill is more prevalent in our innovation networks — on average >50% of users qualify as having discernment skills.

In a way this intuitively makes sense. Creating something entirely new (innovators) is much more challenging than ‘editing’ or somehow enhancing something already existing (discerners).

What are the implications of your results?

What the results now show us is that with this model, we can find people that are noteworthy in certain innovation strengths just from their interaction patterns, and highlight those to our customers, who can then recognize these individuals with particular innovation leadership skills that they may otherwise not have known about.

Furthermore, we are now examining how we can recommend actions to individuals that will best leverage their strengths, in order to improve the overall innovation outcomes in the network. For example, you can imagine that it might be fruitful for someone who is just beginning their innovation journey to get some feedback and review on their ideas from a strong innovator, to help nurture and form the idea into a successful one. By knowing who may need some help, and who would be able to offer it, we can connect people to get even more positive outcomes from your innovation process.

We are also now beginning to understand how a strong innovator or discerner evolves over time — what action patterns are they taking over their weeks and months of participation that build and strengthen their skills. Knowing this, we will then be able to suggest actions to others in order to strengthen their own skills.

Are there additional indicators or results on this you’d like to leave us with?

One interesting view of this work that we’ve been generating is to look at the visual patterns of an innovator’s and discerner’s immediate network — to further gut-check the model and see what a ‘typical’ innovator’s network looks like.

In the illustration below, you can see two different ‘innovators’ in one of our networks.  The one on the left is not yet particularly strong in the innovation arena. The blue nodes are the innovator’s ideas, the pink node is the innovator, and the green/red nodes are the votes from others thus far on the innovator’s ideas.

The left innovator has one idea that is starting to take off, but many additional ideas that are not yet doing well in the system.

The innovator on the right, on the other hand, has a large body of ideas, all of which thus far are gaining good traction in the system.

Screen Shot 2013-10-28 at 12.07.53 PM

We’ve been using these visualizations to help us quickly validate that the model is characterizing individuals in a sensible way. But it also will likely be useful for individuals to eventually be able to ‘see’ their visual innovation (and discernment) profile and how it compares to the average profile, to help pinpoint areas of focus to ensure continued progress and achievement.

All of this supports the generation of even more positive outcomes from your innovation process, engaging the right people at the right time, and suggesting the most fruitful actions that can be taken next.

Stay tuned for the next topic in our Science of Innovation series, which will cover the characteristics and trends of innovative companies.


Chief Scientist Lisa Purvis came to Mindjet from Spigit. Her team explores ways to measure and model innovation networks of people and ideas in ways that will get the best outcomes for our customers. She holds a PhD in Computer Science from the University of Connecticut, in the area of intelligent reasoning and problem solving. Lisa has 11 issued patents and 15+ published papers.