I spoke with David Pugh, Program Director, Offering Management, Watson Marketing for IBM about Marketing Insights and the audience insights feature. Consider this: you are tracking a ton of data on your customers and prospects, way too much to analyze manually, but also way too much to run basic analytics and get any valuable insights to do something with. Although we know how important data is, there can be too much, and you end up overwhelmed with it.
IBM Watson Marketing Insights is designed to ease that stress for the marketer. Not the data scientist, but the marketer directly. This is a tool that doesn’t require you to understand complex data analysis; it requires you to act on the analysis Watson provides.
Using Watson’s cognitive capabilities, Audience Insights analyzes customer data learning more about customers over time, including their interactions and changes in those interactions. The data comes from web traffic, email, social, purchases online or in store from POS systems. It also includes demographic data and unstructured content such as chat, call center logs and email.
This data continually comes into the system, so Watson is always analyzing the data in near real-time and forming hypotheses.
Pugh said it’s the Audience Insights feature that is particularly useful for marketers. Customers are grouped into audiences with similar customer behavior. The marketer is shown these audiences, with an easy to understand description of why the audience is important. The system displays audiences by impact to the business. An example, there is a group of customers who are highly likely to leave a company, but they are very valuable customers, so the company needs to take immediate action. Here's a sample screen:
A marketer can select an audience and export it to the IBM Marketing Cloud or other application and act. You can also push the audience list to IBM’s real-time personalization engine using UBX.
Pugh told me that this first release of the Audience Insights component of the Marketing Insights solution focused on a key set of use cases around customer attrition, engagement and customer lifetime value (CLV), but they will add more use cases in future releases.
Static audience segments don’t work
Pugh said that static audience segments don’t work and he’s right. Customers change over time. They buy different things, go through lifestyle changes that affect how they think, feel and purchase, they may be become dissatisfied or unhappy and consider leaving.
If you lump a customer into an audience based on certain attributes or engagement, and then continually interact with them in the same way, the potential to miss the mark with a customer is huge. Or if you group customers based on one set of attributes, ignoring other key interactions, you may be interacting with a customer in the wrong.
Maybe you can’t do true 1-1 marketing due to cost, resources, and complexity. But a solution that can continually re-segment your customers based on what’s happening with them right now, well that’s fairly close to a 1-1 relationship.
A fresh perspective on insights
If you are an IBM marketing customer or thinking of becoming one, Watson Marketing Insights should be on your martech stack list. Here’s why I like it. It explains why something is important.
Often, marketing analytics will give you the data, they’ll map out key points of interest, but you still need a marketing analyst or a data scientist to help you make sense of that analysis. It can be complex stuff, and not all marketers have that analytical mindset. This solution explains why the audience was defined and why it’s important to you. And you can see the data there to back the hypothesis up.
The next step to that is to automatically take action and do something like run an email campaign or adjust how the website is personalized, or maybe send an SMS or in-app notification. Watson Marketing Insights doesn’t do that, but Pugh acknowledged that is where analytics is headed. The key is whether the marketer will be ready to relinquish control.
For now, as Pugh pointed out, this is a relationship between man and machine that helps marketers improve customer experience.
I like that Audience Insights can monitor interactions in real-time and continually return updated audience segments, so you are always interacting with a customer for the right reasons. But I’d like to see the next step come sooner than later.
If Albert can figure out the best way to interact with an audience and adjust those interactions on the fly, I think Watson could likely figure that out pretty quickly too.
Not all marketers will be ready to give this much control to a machine. Those that do will have a lot more time to focus on the bigger customer strategy work, including improving interactions across the complete customer lifecycle - you know, that work they say they want to do but can’t get away from the day-to-day business to do it.