Do you want to know how to bring down the mood at the next happy hour, cocktail party, or cookout that you attend? Bring up predictive analytics.
In reality, predictive analytics is not scary at all. Here is a great explanation from Dr. Eric Siegel, founder of Predictive Analytics World and author of the new book "Predictive Analytics: The Power to Predict Who Will Click, Buy, Lie, or Die."
Predictive analytics is the use of historical performance data under specific conditions to predict a future result with a high degree of certainty so that you can take actions now to affect those results.
What does that mean in the real world?
Take your customers in your online customer community for a example. Assume that 80% of your customers continue doing business with you year over year. If you learned that of the customers who accessed your online customer community once a month, 85% of those customers renewed their contracts with your company. Then, you learned that you retain 97% of the customers who asked a question, started a discussion, or responded to a thread in your online customer community.
Using the data in this simplified example, you can predict that your average customer retention rate will increase if you can drive more of your customers to participate in your online customer community. That basic predictive model combines demographic, transactional, and social/behavioral data to help you make smarter business decisions.
If you take the profile of someone who exhibits a behavior that you want customers to take, you can overlay that on top of current and future customer data to identify which customers are more likely to take the action that you want them to take. This is the core of social crm (customer relationships management).
Putting aside your unpopularity at most social gatherings, predictive analytics is very exciting. Applied to any part of life or business, predictive data is the closest you will get to a crystal ball. Imagine the possibilities 'business strategies, elections, sports, college enrollment, and investment outcomes. Predictive models are being used in all of these environments today' and they are being use to generate winners.
Social networks have accelerated the usefulness and accessibility of predictive data for most organizations. Whereas, predictive models used to rely mainly on in-person, event, phone, email, and anonymous website interactions to compile a behavioral profile of specific audiences, social networks offer another dimension of peer-to-company and peer-to-peer activity. Social data enables companies to micro target with more accuracy using many-to-many conversations, rather than just linear person-to-company communication.
However, tracking and analyzing the entirety of your ecosystem (customer, members, prospects, advocates, partners, etc.) on the social web can feel like trying to swallow the ocean. There are very sophisticated tools available for processing this kind of information. However, it can take an extensive investment in personell and technology to bring the strategy to fruition.
For many businesses and membership organizations, private online communities present a more rich and manageable environment for capitalizing on predictive modeling. Online customer communities act as a microcosm of the entire internet. The closed environment enables customers to do more, from watching tips and tricks videos to submitting product enhancements to participating in forum and listserv discussions.
While your target audience has more ways to share ideas, get support, and provide feedback, private social networks offer your organization better tracking of the social activity compared with monitoring public social networks.
In this example, we'll look again at customer retention, a popular business intelligence model that uses predictive analytics. By examining your online customer community data, you can create a profile (demographic and behavioral) of the customers that stopped doing business with you (transactional) last year. By also looking at the social activity of customers that did renew or make additional purchases, you can determine which social activities in your customer community were done or not done exclusively by customers that you lost.
Let's assume that the following three behaviors were in the profiles of customers that were lost:
Next, identify all of the current customers that fit that profile. There is a high degree of probability that these customers are at risk for ending their relationship with you.
Identifying customers that may need special attention to keep them as customers is just one example of how organizations use predictive analytics from their online customer or member communities. Other models include:
These are just some common scenarios. If a specific audience uses your private online community, you can design a predictive analytics model for getting better results in practically any business strategy.
Which behaviors do you want to predict? It is important to start with the outcomes that you would like to predict. The result of this initial discovery process gives you the questions that you are going to ask into your "crystal ball." It is the foundation for the next three steps.
During this step, you'll also want to determine the actions that you can take once you identify a specific segment of your ecosystem. For instance, if you are looking for customers that you would like to partner with to expand your customer reference program, lay out a preliminary roadmap for how you will approach the target list of customers.
Which demographic information, transactional data, and social activity were common and unique to people who exhibited that behavior? In this step, you'll examine the common characteristics and activity of those people who have, in the past, demonstrated the behavior which you are after.
In the customer reference program example, you might use your online community software to research the social activity of your best known customer advocates. Did they answer questions posed by other customers in the discussion forums? Do they maintain a blog in your community? Do they access the community several times a week?
This is a very important step. Since your customer, member, or prospect profiles layout the behavioral roadmap that you base the other steps on, this step deserve extra rigorous analysis. If you don't correlate your target audience and outcomes correctly, you may end up pinpointing and promoting an activity that has no bearing on the results you want.
Take the profile you set up in the previous step and overlay it onto existing customers, members, prospects, etc. In this step, you'll generate lists of online community members that fit that profile. Using your online community software platform, you can group members of the community according to specific social, transaction, or demographic data.
If you are seeking to identify new potential customer advocates, compile a list of current customers that meet the social, transactional, or demographic criteria that you established in step 2.
This process will leave you with two buckets of online community members - people who fit the profile and people who don't fit the profile. Each of these lists can then be analyzed further to discern degrees of strength in meeting a specific profile's social, transactional, and demographic criteria.
Now, you have these two groups of people, it is time to take action. For those prospects, customers, or members that fit one of the profiles you set up, it is time to follow through on the action plan you developed in step #1.
This might mean that you launch campaigns to reach out to "at-risk" customers. This might mean that you target your event marketing toward customers or members that are more likely to register and attend. It might mean that you have your next recruitment class for your customer reference program.
Sometimes, these are positive actions, like further engaging customer advocates. Other times, your action plan will consist of proactive steps to prevent bad things from happening, such as losing sales, customers, or partners.
In addition to personally reaching out to people if necessary, this process also involves adjusting your community management processes. Set up paths for each of these groups to take to move from one bucket to the other.
For an example, if you learn that customers that access into your online customer community once a month buy twice as much, establish paths that customers can take to access your online customer community more often. This could be a combination of customer experience, usability, marketing campaigns, and one-on-one outreach. The goal is to make it as easy as possible to move from one bucket (customers who do not access your online community at least once a month) to the other bucket (customers that do access your online community at east once a month).
Strategies behind online customer and member communities are shifting rapidly. While an active online community is still a major differentiator, what you do with the data in your community is becoming the most important element of your social crm strategy.
There are endless ways to use predictive analytics in customer acquisition, retention, and engagement. As you read through the steps in the article, you'll notice that the main component in data. You must have the data to be able to build your organization's predictive models.
Most online community software has the ability to capture behavioral data. In you don't yet have your online customer community set up, select a platform that has enough features to allow you to use data to delineate between those members that exhibit the behaviors you want and those that don't. It may be difficult to tell the difference with more basic online community systems. According to Dr. Eric Siegel, if you have more data, you have more opportunities to learn.