Creating a Next Best Action Strategy with AI, Part 1: Next Best Offer

Using the appropriate types of data at the right time in the customer journey is essential to modern marketers. While being able to create an effective buying pathway that can work for your entire audience is important, it is increasingly critical that marketers have an understanding of providing a next best action that is unique to the audience segment, if not the specific individual.

We’re going to explore next best action and how it can play a number of different roles for organizations over the next few articles, because it is not solely relegated to the buying process either.

But first, for those of you less familiar, let’s define next best action and why it makes sense to take this approach. While traditional marketing and even more recent digital marketing has been driven by what the marketer wants a customer to do, next best action is centered around what the customer wants. Put another way, it’s a design thinking approach to marketing and the customer experience.

Over the next few blog posts, we’re going to explore a few ways that next best action can incorporated as a strategy to reach customers and provide a meaningful experience.

Next Best Offer

An immediate first place to start in creating a next best action strategy is to utilize what is often referred to as “next best offer.” This means that, based on a user’s behavior and other information you can determine about them, you provide them with the most appropriate incentive to take an action.

Traditionally, marketers have made assumptions that all customers are willing to take the same steps in order to complete a desired action, such as buying a product online, signing up for a checking account, renting an apartment, or any other purchasing behavior that takes multiple steps. Thus, the buyer’s journey looks the same for all customers, as shown below:

NBA-comparison-1st.png

But the above scenario doesn’t take into account any of the types of information we might have about a customer that allows us to better understand them, segment them into sub-groups, or even better yet, personalize an experience that works best for them.

Using AI and machine learning, you can learn how different types of customers behave, what their buying preferences are, and the types of channels, messages, and incentives required to move them to a purchase decision and action. When you begin to do this, you will quickly see how what might seem like a single audience segment is actually a lot of individual behaviors. This can help you create sub-segments.

In the figure below, you can see how three different unique customers (or customer sub-segments) take three very different buying behaviors. The different colored rectangles represent different channels or communications (e.g. Search ads, email, social media, etc.). Each customer has different ways to reach the purchase completion step, and not all even take the same number of steps.

NBA-comparison-2nd.png

In the scenario above, notice that “Customer 2” only took 1 step to convert to a buyer. How much more efficient is it for a company to know and understand when a certain customer type needs less pushing than others. That could translate into literally saving money, by forgoing expensive advertising targeted at those users, or it could simply cut down the buying cycle so you capture that revenue more quickly.

The Definition of “Best”

Keep in mind that the best offer is not always the cheapest price, or lowest rates. Remember also that a “best offer” should be the best one for both the customer and the company. For instance, if you have attracted a customer who may be likely to purchase, but does not fit the profile of a repeat buyer, it may not be worth getting them in the door with a massive discount. It may be a better use of your marketing dollars to find and target other customers who will have a longer lifetime value, and simply offer the low-value customer a moderate incentive and take the risk they may leave.

In order to determine these things, you need to create models to determine an expected value for each customer, as well as a “risk score” that calculates the probability that the person will either abandon the funnel during the sales process, or simply be a one-time customer that doesn’t translate into a repeat buyer.

In the next article in this series, we’re going to go beyond the next best offer, and see how a next best action strategy can benefit education, creating long-term customers and repeat buyers.