How AI and Machine Learning May Apply to Marketing
Most of the talk about current use of artificial intelligence (AI) in marketing cites what the 4 Horsemen are doing or likely doing. Google, Facebook, Apple and Amazon have digital, data and automation at the heart of their businesses. They have the skills and the culture to make the application of machine learning and artificial intelligence a core competency.
Meanwhile, many marketers still struggle just to get their customer data source in order never mind speculating on how AI might improve the customer experience and/or the profitability of these customers. The hype curve of AI creates this nervous energy for marketers. To not do anything in AI seems wrong. On the other hand, to leap in expecting it to solve big, hairy 2018 problems also seems naïve.
I would recommend creating a simple ‘road map of exploration.’ Andy Betts (@andybetts1) has a useful post and diagram (above) counselling CMO’s to start planning in 2018. For once, it’s okay to start learning and exploring vs. jumping into action. I have a ‘bias-for-action’ like most marketers. We want to learn by doing. In this case, I want to pick and choose the pilots and proof-of-concepts (PoCs) while not derailing my marketing and customer service operations who already have a pretty full agenda next year. I will start by defining use cases where AI may have a positive effect on marketing.
Five Use Cases for AI in Marketing
Show the right, next call-to-action – Are people looking for a home-plus-valuables insurance quote? Trying to get answers about coverages in the weeks leading to an insurance need? Winterizing their home before jetting off to Florida for the season? These are pretty simple states but often hard to detect amongst the unknown visitors to your digital properties. Knowing the next right action to suggest and guide them is hard.
Propensity modeling used to be the answer. Defined as “a statistical scorecard that is used to predict the behaviour of your customer or prospect base,”propensity modeling works best when you know a bunch about the person. Think rich customer databases vs. a few Web clicks from strangers on your Web site.
Would you rather have a “scorecard” or a really smart robot (i.e. AI)? Imagine paying attention to the patterns of visitors to your site and all of the content they consume. Imagine if you could reliably predict/understand what they are trying to accomplish and you were able to deliver just the right calls-to-action on their screen. No more junking up their view with every possible choice they may consider. No more agonizing over Web site navigation models. With AI, you could understand what they are after and just give it to them.
Identify more accurate and actionable “look-a-likes” – Finding more prospects that look like your most valued customers and then selling them your product is generally a good thing. Most of the social networks have offered look-a-like audiences for a few years now.
