Would you like a powerful alternative to traditional push marketing? One that helps your customers find the information they want in the channel they want at the time they want it?
If you’re like most marketers, your answer is, “Yes!”
You can think of Next Best Action Modeling as a customer-centric approach that helps your customers take the “next best action” after they complete one action. It’s similar to Amazon’s recommendations – since you bought x, you may also be interested in y.
When done well, it’s incredibly powerful. When not done as well, it’s like the example I gave in a previous article about how I searched for an automatic pet feeder and bought one. Then for the next 6 months it felt like every site I went to offered up ads on automatic pet feeders. What a waste of customer insight. Rather than flogging a dead horse they could have used that information to offer up ads for other relevant pet related products, at the time I was in a buying mode. For example, while I am doing online grocery shopping, they could have offered up pet food offers, or flea treatments in summer and so on. These would have been welcome reminders and could have resulted in a sale rather than being an annoyance. Instead, they wasted advertising dollars by offering me an item I’d already purchased. All because they had SOME customer insight – I have a pet – but had not used it effectively.
Next best action modeling changes the paradigm. Instead of finding the next customer, it focuses on finding the best proposition for the customer to add value to them and be relevant for what they need at that time.
This is a customer rather than product focused approach and it treats each customer as an individual person within the context of their individual behaviors (past and present), needs and preferences.
The Cornerstone of Next Best Action (NBA) Modeling
You may be thinking NBA modeling makes perfect sense. You’ve already experienced it with Amazon, Netflix and other business models, so you know it works.
It also works in the pharma industry. In the past six months, Eularis has used NBA as part of several projects. The foundation of NBA is a combination of data and advanced AI (artificial intelligence) modeling to ensure that the most likely outcome is predicted from a set of interactions with a customer or customer segment. To be successful, it needs to be well-planned.
To start, it is useful to create a strategic business case, based on what data is available. Then we conduct a thorough analysis of all data. We use AI to do this (as it is the only way to analyze all the data effectively), to plan customer actions and communications for every channel and line of business and – particularly in the case of inbound contact – it must be possible in a matter of seconds.
Once you have analyzed the data, segmented your customers appropriately, and planned your business cases, then you can create a decision engine powered by AI. This decision engine calculates the next best action based on what the person has done in the past, what others with very similar attributes have done, and the results, and what they are doing on your web site or other property right now.
Whatever the touch point – be it a website, a call centre, a CRM system, the information can go directly into the decision engine where it will be analyzed instantly to come up with the next best action recommendation for the customer. This makes all channels consistent and work together in an integrated fashion.
However, as in my pet feeder example, the critical thing is to make these actions and content relevant and appropriate for that moment. The appropriate content may be a thank you message or a services action or something else rather than a sales focused message. You do not want to be selling when selling is not appropriate. You want the customer to come back when they are ready to buy.
Here’s a Step-by-Step Process of Creating an NBA Model
1. Create dynamic and ongoing segmentation of your customers into micro-segments who are predicted to behave in the most similar way to various content.
2. Model the behavior to predict how each micro-segment will respond to marketing actions.
3. Predict impact on market share or sales over a longer term (as the approach is not a hard sell or direct sales approach).
4. Ensure you use an AI approach that is self learning such as machine learning to constantly track and test and optimize based on all new customer actions being tracked.
Of course, even if you have access to all the relevant data, creating this system is fairly complex and takes time and expertise.
