You may be asking yourself, “What’s the idea behind predictive marketing, and how would it benefit my business?” To put it simply, predictive marketing is the use of user data and historical user behavior to anticipate future behavior. That might sound like a little bit of hocus-pocus, but if the concept is applied correctly, it can provide new and interesting insights into your customer that can lead to more revenue and better results. And the big secret is that many companies are already doing it. You should be, too.
The Key Ingredient: Data
Before we can jump right into the process of predictive marketing, we need to understand the major requirement for predictive marketing: a strong amount of data. The results of this endeavor are highly correlated to the amount and accuracy of your data. The simple adage of what you put in is what you get out holds true in predictive marketing. Imagine trying to guess someone’s shoe size based on his or her driving habits. Although you might get lucky, having more data on the person would increase your chances of making an accurate prediction. So when you think about predictive marketing, remember this: the more data you have, the more accurate your predictions will be.
What Data Do I Need for Predictive Marketing?
To start doing predictive marketing, you can have as few as 10 or as many as a thousand data points about your customer. The most important ones you need to know for your customer are demographic and behavioral data. Some examples of this data would be:
- Website analytics
- Media channel usage
- Transactional data such as purchases or store visit,
- Demographic data such as age median income, ZIP Code, etc.
Building Cohorts in the Data
If you break the data into the two groups of demographic and behavioral data, the demographic data for predictive marketing allows you to be able to assemble cohorts. Cohorts allow you to increase the accuracy of your insights and prediction, and give you a reasonable target for marketing experimentations. For example, if we have correlation between a certain age, race, sex, and the median income of the customers in a particular ZIP Code, we could perform some predictive analysis for all customers to see if they share similar behaviors. In the same sense, when we don’t have that kind of demographic data, correlation with peers becomes increasingly difficult to link behaviors across the board with customers.
You can also group by behavioral data, but without demographic data you’ll experience wide variations that can only be explained by demographic limiters.
When it comes to behavioral data, the main requirement is that we need behavioral data over an extended period of time. Correlation is strongest when it can link together behaviors seen in relation to other behaviors, which need time to develop. For example, if you only have three interaction points that exist within a month, your window of visibility to what can happen next is very small. Yet, on the other hand, if you have three years of data, it becomes much easier to see what the next month or two months may hold for a particular customer based on repeated behaviors.
How to Manage Predictive Marketing Data
The end goal with all of this is to execute marketing campaigns, so it makes sense to select your CRM and marketing automation tool first. That way, you can see what kind of predictive workflows and data analysis you can create in that software before buying more software. If you have more in-depth data to start, you may need to consider data analysis tools such as bigML and Tableau Public, which can help you assemble this data outside of the marketing context. This might be a great conversation for your business analysis team or IT department, who can help you acquire and deploy data tools to inform your marketing efforts.
All of this might sound like an uphill battle, but don’t get discouraged. Even if you’re using Excel, you can still get some really strong insights without all the fancy tools.
Next Steps
In the second part of this post, we’ll talk about taking the data and performing predictive analysis based on the behaviors that can be seen. And in the third post, we’ll talk about using those predictions to test and use marketing automation to its full glory.
