Tim Ebner is communications director and press secretary at the American Forest & Paper Association in Washington, DC. He is a member of ASAE’s Communication Professionals Advisory Council and a former Associations Now senior editor.
Big data, data analytics, and artificial intelligence are enabling predictive analytics used to anticipate needs, opportunities, and threats in an organization’s environment.
They may not know the term, but most American consumers are already familiar with predictive analytics. This data-driven discipline enables companies like Amazon, Netflix, and Spotify to offer individual users recommendations for a product, show, or song based on previous purchases or engagements. And that experience in the marketplace is creating new expectations for how associations serve their members.
“These companies are the gold standard. That’s really what associations are up against,” says Thad Lurie, CAE, vice president of business intelligence and performance at Experient, an event management company. “Their entire business proposition is based on the ability to predict what you will want.”
Current estimates put the predictive analytics market on a rapid growth trajectory: It’s expected to become a $4 billion to $9 billion global market by 2020, according to various estimates. This new way of doing business is causing many associations to rethink their data strategy, Lurie says.
“To prepare for this, you need to get your data house in order,” he says. “You need your data to be consolidated and clean, and the more data you have, the more accurate you will be.”
Sure, but creating a version of the Amazon model can seem like a pipe dream for associations with small tech teams and shoestring budgets.
“The key here is not to get discouraged,” says Laura Lewis, director of technology at the Society of Critical Care Medicine (SCCM). “Sometimes at associations, people are afraid to make a decision, and you don’t move quick enough to take advantage of the data trend. We are agile enough that, yes, we sometimes make mistakes, but we can also experiment and learn from those mistakes.”
This type of experimentation typically starts with a simple question or hypothesis. For SCCM, it was this: What actions make people feel a part of our community? By answering that question with historical data, Lewis and her colleagues could begin to see how a member moves from engaged member to volunteer leader.
SCCM has about eight years of member engagement data stored in its association management system. For every action a member takes—paying dues, writing a journal article, purchasing a product, or serving on a volunteer committee, for example—a point value is assigned. A member’s accumulated points then translate to an engagement score.
That’s a useful measure of participation, but it doesn’t really speak to what makes a volunteer leader tick. To get that answer, Lewis and her team had to dig further into smaller data sets.
“We took our council members and looked at the activities that they did as they moved through their career stages,” she says. “We started digging and looking at things that brought us to specific behaviors, and we started to see patterns—they joined certain committees, [or] they come from this kind of institution.”
Lewis and her team found that certain actions, especially at the earliest stages of student membership, could help predict who might become a volunteer leader in the future. That intelligence is helping SCCM to think strategically about how to grow a pipeline of leadership talent.
Her advice for associations just getting started with this type of data analysis?
“Take your data in small bits, and don’t just think you can plug things into the computer, and it will spit out the perfect answer,” Lewis says. “You’re going to go through iterations. You can’t be afraid to try things, and you have to be ready for failure, because a lot of this is trial and error.