Data Analysis Should Be a Social Event
Here's a common business problem: You want to retain more of your customers. Depending on your corporate analytics culture and the experience of your analytical team, your analysts will probably approach this problem by asking one of the following questions:
- How do we create the best model to predict which customers are most likely to churn? In this approach, analysts treat customer retention as a highly structured problem to be solved formally. The search for the solution is about finding data and the best algorithms to apply to the data. In our experience working with hundreds of companies, this is a very common approach but it is essentially a tactical one. Big data managed this way may well improve your retention numbers but not transform your company.
- How can I help marketing retain more customers? More experienced analysts will take a broader view of the retention issue. Rather than framing the challenge as one of building a better churn model, they will look to improve their understanding of the customers behind the data. What kinds of products or types of communication are going to increase customer loyalty? This approach to data analytics has the potential to usher in bigger changes than the first approach but the changes will remain relatively incremental. The company may try a new communication channel, new packaging, and perhaps even change features of the product itself. But that will be about as far as it goes.
Both approaches clearly deliver value. However, we've found that companies that don't explore the social aspects of analytics are missing out on opportunities to use data to completely transform their businesses. To be sure, much has been written about analysts as storytellers (see, for instance, A Data Scientist's Real Job on HBR.org) and about creativity as part of the analytical process, but what has not been explored is the storytelling analyst actively participating in co-creating value with others.
Psychologists believe that creativity flourishes in social contexts, as thoughts are translated into words, objects or images and in turn reformulated into ideas (this is one of the reasons why visualization of data is so valuable). On the management side, there is increasing evidence that co-creative processes involving consumers and other stakeholders can have a transformative effect on key processes such as new product development.
Here's how it works in data analysis. You put together ad-hoc teams that involve not only analysts with relevant domain expertise but also represent skills from other domains, which brings new ways of thinking to old analytical problems. Over a short two or three day period the team will brainstorm around the problem involved and bring together as much data and as many analytical frameworks as they can to both frame up the problem and outline potential solutions or at least pathways to solutions.
A compelling example of the co-creative analysis process can be seen at DataKind, a charity devoted to helping other charities extract value from their data. What DataKind does is bring together analysts, domain experts, and anyone with a passion about the problem in question for weekend "datadives" to "analyze Big Data in the service of humanity. In March this year, for example, it hosted a two-day session in co-operation with the World Bank, various UN agencies, and the Qatar Computing Research Institute to study data around poverty and corruption.
This approach works well in business. We recently participated in a data-dive type process at a major telecommunications company. Like many telcos this company is trying to woo customers through customized marketing communications and offers. A team made up of analysts and business intelligence people from the operator plus some external consultants (ourselves) identified and applied an analytical methodology from the retail industry to quickly assess the needs and preferences of thousands of actual and potential customers. They built on these assessments to create 30 finely targeted marketing campaigns. Two-thirds of these were successful beyond the company's wildest hopes. All this came from just three days of group brainstorming.
As with other co-creative processes, co-creative analytics comes with risks that must be mitigated through providing strong leadership and specifying clear deadlines and outcomes. But the fruits of the creative energy that you can unleash through teaming analysts and domain experts in these ways more than justify the investment of time and money involved.