Interaction Optimization vs. AI: A Three-part Series
Part 2 of 3
How AI Can Help Solve Three of the Biggest Problems Facing Digital Marketers
In the last blog, we discussed the buzz coming out of martech about AI and took a little trip down memory lane to revisit early interaction optimization tech like heat maps and A/B tests and some of the thrills the related experimentation gave us. What we realized in the process is that we’re still chasing the same problems. It’s just that now we have something better, AI and ML, to help us solve them.
So, what are the problems a machine can solve better than man, or in our specific case, a digital marketer? For me, it all boils down to three things: (1) Speed & Efficiency, (2) Decision-making, and (3) Continuous improvement.
Let’s look at each one separately.
1. Speed & Efficiency
Let’s begin with speed & efficiency, and the simple fact that machines can process information at speeds infinitely faster than a human. If the problem is how you test and improve something before the question becomes irrelevant…there’s no competition – the machine will always win.
There’s an example of this we like to share. It involves one of our retail clients in the entertainment media space. The rumored death of a superstar recording artist had just hit the Internet, spiking interest in his music across our client’s site. With the use of AI and ML, the client was able to quickly pick up on the trend and use our technology to test and prioritize placements to satisfy customers’ need to commemorate him. The temporality of a hit song or movie brings the need for a solution into stark relief.
Going back to my consulting days at Blockbuster, I remember the imperative to anticipate trends, so we could shape demand and maximize every revenue opportunity was ever present. It still is. Now we just have a better way to solve it.
2. Smarter Decision-making
That brings me to the second problem of how you make your decision-making smarter, as well as more timely. Even without AI, we have the ability to test, create rules and segments, and select winning treatments that effectively operate as decisioning for our audiences. But it’s years behind what we can accomplish by machine. With AI and ML, you don’t just get the granularity and the ability to detect and recognize purchase intent but you can now also act on it and make smarter analytical decision in-stream, and in real-time. It’s just far more than we could ever do by running it, for example, as a rule-based treatment against a segment for a particular day part and product grouping.
3. Continuous Performance Improvement
The third problem where AI and ML can help is the thorny one of continuous performance improvement. Up to now, we’ve been stuck with discrete, time-bound experiments, which are difficult to manage and optimize for flow, especially with respect to campaigns. That is where it gets interesting because it’s something that even AI can’t solve alone. There are plenty of people today throwing fixed algorithms at it. Maybe they’re rescoring the model every day, based on retention or propensity, but they’re still not leveraging AI’s incredible potential to learn and improve. It’s only when you add the self-learning and real-time components that you start to open yourself to the real possibilities of the technology. It’s here where you can start to create new models on the fly and adjust to shoppers’ tendencies as they reveal them. It lets you understand and learn from the variability in shopper behavior from morning to evening or season to season and rejigger the models immediately to act on it.
Connecting the Dots Across the Customer Journey and Enterprise
Look, we live in this incredibly exciting time, where we’re creating all this data that if even partially harnessed could tell us so much about our business and customers – and ultimately force us to rethink our approach to everything. Before when we were just doing site or interaction optimization, the focus was on our logs and what was occurring on the site or at the touchpoint. But now, with AI and personalization AI, we can finally begin to bridge the gap, bring in and combine all this data – CRM, financial, social, and operational – and connect the dots both in terms of business objectives and the entire customer experience.
Data and instrumentation give us the ability to detect things and act reflexively to exploit them. But it’s the AI – or more intelligent AI – that can take the next step and, for example, apply what’s going on with a supply chain to shape demand, so we can maintain the right velocity, gain efficiencies, and drive higher margins and profits. Connecting the dots, however, isn’t just about making smarter business decisions; it’s also about understanding the lifecycle of the customer and how to optimize for it. So in the above example, we’re not just promoting an item, like printer toner with excess stock, to everyone but, instead, are targeting customers that we know are in need of replenishment.
So now that we’ve established all the problems machines can solve by themselves, what does that leave for the human marketer?
We will examine this in the next blog on how machines make humans better, and vice versa.