Posted on

Artificial Intelligence – The Next Marketing Frontier (and Danger)

Artificial Intelligence seems to be everywhere all of a sudden, and it’s making its way into the technology marketers use every day.  We’ve been talking to (with?) our smartphones for a while now, and there’s pretty much no mobile device that isn’t just begging to hear your voice.  With Apple adding Siri to the Mac, that now applies to pretty much any device. Amazon even has an eerily lighted cylindrical device that can play Jeopardy! with you―and presumably understand your stuttering, uncertain responses in the form of a question.

The Direction of Artificial Intelligence Operational Systems

It’s only in the past year or two that we’ve seen so-called artificial intelligence technologies make their way into operational systems and software used in business.  And now we’re seeing some form of intelligent capability make its way into marketing.  Here are a few examples:

  • Ad Targeting: Machine learning is working for companies such as Baidu to determine when any given user is most likely to click on what kind of ad, then automatically serving the right kind of ads for that user at that time.  This attempts to maximize click-throughs.
  • Recommendation Engines: We’re all familiar with the jokes about Amazon’s (and other vendors’) “people who bought this item also bought” feature which often make less-than-ideal recommendations.  Applying machine learning to large amounts of data on consumer behavior, however, can improve this dramatically.  Under Armour is using IBM’s Watson to analyze its own customer purchase data with third-party data on fitness and nutrition to serve up far more relevant product recommendations.
  • Preventing Credit Fraud: Banks have been using massive amounts of data to try to determine when a particular credit card transaction has a good chance of being fraudulent.  Now companies such as USAA are using natural language processing algorithms to look at the text within transactions to determine potential fraud even without a previous pattern having developed.

These examples are taken from a pretty interesting list of applications of artificial intelligence in marketing.

The key question for me is:  How will this change marketing?  I’ll offer some thoughts on where this is going, but first indulge me a short background explanation.

There are two main lines of thought in the computer science world about how intelligent systems (from software to robots and beyond) should act and interact with humans.  One says we should be developing systems that are independently intelligent.  Those systems would learn from the initial set of experiences humans provide but then would function on their own, without human guidance and making their own decisions.  If this scares you a bit ―and it should―you can read an incredibly insightful speculation on this in Asimov’s classic, I, Robot.

The other line of thought says intelligent systems should be built to extend and enhance human intelligence.  Sometimes this is called the “Star Trek” school of thought since that is how the computer systems on that show generally operated (and the independent androids were almost always evil―apologies to Mr. Data). The goal here is to help humans advance their own thinking.

A good example of the latter in the marketing world (and a function I really hope to see one day soon!) is the ability to ask your marketing automation systems questions such as “What are the top three paths people who buy our products take through our website?” and have your system know it needs to crunch the behavioral data to develop paths and determine outcomes.  Even better, the system would know why.  In this case, the system is not making decisions on website structure or how to present what information to whom, but it is telling you, the human decision maker, what you need to know to make those decisions.

Most of the examples of intelligence, including machine learning and natural language processing, that are in place today fall into the latter category:  they exist to provide some form of information plus analysis to a human decision maker.  There are a few examples of systems that are given jobs they do themselves (such as the ad server example above), but even those are assigned a specific job and decision-making framework by the humans who control them.

What Comes Next?

I think the next few years will bring a dramatic increase in the intelligent capabilities of all kinds that will be brought into business systems, including marketing systems.  Nearly all these will fall into the data or language analysis category at first.  They will do things such as answer customer-service requests or help marketers make sense of large sets of unrelated data.

But some will start to make some of the decisions marketers make every day.  For example, IBM’s Watson technology analyzed millions of recipes and now can develop a (presumably tasty) recipe given a set of ingredients.  That’s why they let it compete on Jeopardy! but not on Chopped!  Imagine if Watson’s artificial intelligence analyzed the marketing mix of every company in your segment and added in the consumer behavior data.  I’m willing to bet it would make marketing mix and timing decisions as well as any of us could―maybe better.

In the retail business, it’s not hard to envision the day when Apple’s intelligent ear pieces (version one comes out later this month in the form of AirPods) can remember that three days ago you mentioned you needed to buy new socks, and then remind you (literally putting a “bug” in your ear) as you walk by a sock-selling retailer that is doing location-based advertising with Apple.  Today this feels intrusive, but as our artificial intelligence assistants become more intelligent, it will seem less so. It’s more like having your friend notice the socks in the window and asking, “Didn’t you mention you need socks?”

Will Marketing Change?

No.  Marketing, as we sometimes need to remind ourselves, is a discipline, not a set of actions.  We’ll still be doing it.  But our jobs will change dramatically.

The predictions for future jobs are dire.  This article predicts Robots (or some form of artificial intelligence systems) will take over 50% of all jobs within 30 years (the career span of someone 10 years out of school!).  And if you Google “Jobs AI is Taking Over,” there is no end of similarly dire predictions.  Many of those jobs are white-collar jobs.

Some of those will be marketing jobs.  As systems get better at analyzing data and behavior, the jobs devoted to that will disappear.  As systems get better at processing and responding to natural-language requests, customer-service jobs will start to go.  And yes, as systems get better at inductive reasoning, jobs that focus on messaging and positioning (inducing target customer desires) will also go.

The flip side of all these dire predictions is that more intelligent systems will help the marketing decision makers make better, more informed, less biased and faster decisions. Those of us who focus on creative roles and on decision-making roles will get much, much better at our jobs, as our systems help us do better, faster.

Now, if Siri could only tell me where I saw that really cool gadget I wanted to buy.  Or even reliably get me directions home.  Then I’d be convinced we’re on our way to more intelligent systems.

Are you using intelligent systems in your marketing?  Are they helping?  I’d love to hear your stories.

[alert color=”white”]

MENG is the indispensable community of executive level marketers who share their passion and expertise to ensure each member’s success.
Learn more — Apply for Membership

[alert color=”white”]

Posted on

Caution: Predictive Analytics May Miss One Important Thing

Predictive analytics is, without a doubt, the new big thing in marketing.  It’s how we marketers are putting so-called big data to work to help us find, target, and sell to the right customers at the right time.  I’ve written about this before; any time we rely on technology or process to tell us about our customers’ preferences, habits, or needs, we run the risk of missing out on one critical element of our customers’ decision-making.  Our customers are human and, therefore, somewhat unpredictable.

Many years ago, I worked with a company that created customized news feeds.  Customers would select their areas of interest, and each day, the company would sort items from a range of newswires (yes, this is before social media!) and send each customer a custom collection of articles, press releases, and other news items.

A general concern others and I raised about the trend toward more personalization (which is still ongoing) is that people would miss out on items of general interest.  In those days, when I read the newspaper, I would seek out sections of particular interest to me, but I would also read the front page and often catch other items on my way to my sections.  This exposed me to news, information, and thinking outside my specific area of interest.  In particular, reading the front page gave me a sense of what was collectively considered important (as filtered by an editor, granted, but one whose interest likely was matching the collective interest).  This provided a common understanding of the world and important events of the day.  With an entirely customized newswire (or, in today’s terms, a group of Facebook friends with whom you completely agree), you create your own unique understanding of the world around you, and you become less aware of what is outside your bubble (and in some cases, less able to understand it).

One of our goals as marketers is to influence behavior, particularly toward buying our products and services.  One way to do this is to create some elements of a bubble around the target buyers so they see more of your offerings than anything else and more messages encouraging the lifestyle associated with your offerings.  That creates stronger associations with the promise of the brand and results in brand loyalty.

We observe the actions customers take and focus on the ones that make them most likely to deepen their association with our brand and all of the things for which that brand stands. That creates the customer journey.

Once we know all that, we work hard to influence customers to take the next step on their journey toward becoming a brand loyalist and buying more and more from us.

Dealing with the myriad actions, possible paths, probable journeys, and the wide range of customer tastes and behaviors has been nearly impossible until technology stepped in to give us ways to store and analyze all that data.  Enter big data and predictive analytics.

We now have computer systems that tell us—if a customer has a certain set of tastes and preferences, and then takes a given action (or a series of actions)—what the most effective way to get them to take the next action is.  So we do that.  Then we see many of those customers taking the hoped-for action.

Enhancing the Impact of Predictive Analytics

One of my favorite themes is to remind marketers that your instinct—your intuitive understanding—goes far beyond the analysis of any computer system.  You will not always be right, but your intuition provides a strong sanity check.

For example, your predictive analytics might suggest prospects who end up buying from you always take a specified action several steps prior to purchasing.  This might be true.  But you might also notice there is a large drop-off rate right before that step—only a small number of prospects in your funnel move to that step.  Your analytics don’t tell you this because you’ve told your systems to answer the question of what causes people to buy, so it looks at the outcome and works backward.  It takes human powers of observation to look at the funnel from a different perspective.

I rely on my systems and the analyses they produce to tell me how my programs are working.  I set them up to give me data-driven answers to a variety of questions, including the question of what actions are most influential in converting prospects to paying customers.

But I always look at the data myself.  I look for anomalies.  I look for things that might not be answered by my systems the way I’ve set them up.  I look for things I’ve otherwise overlooked.  Some of those insights have led to opportunities I would not have seen otherwise.

In short, I use my own experienced intuition to make the final call about what is working, what is not, and where I should look next to improve my efforts.

Don’t miss out on this one critical factor.  Don’t let your predictive analytics and automation systems take over your marketing.  There may come a day when intelligent systems can do this for us, but for now, this is your job.  It’s where your value gets added.  Using your own experienced intuition is what makes the difference between good marketing and great marketing.  Don’t give that up.

[alert color=”white”]

MENG is the indispensable community of executive level marketers who share their passion and expertise to ensure each member’s success.
Learn more — Apply for Membership

[alert color=”white”]