Provocateur

Algorithms Can’t Hug

By Nick Godfrey  9.10.18

Algorithms have been around for a long time. Since about 2000 BC long. But where algorithms were once just the focus of mathematicians, scientists, and engineers, they’ve now become an integral part of our modern marketing conversation.

 

The problem, though, is that in many cases these conversations have become more about how many algorithms we have and less about what they’re doing to enhance the customer experience.

 

Wherever there is a need to predict, automate, and respond, there is—or can be—an algorithm. AI, big data, and our growing focus on automation to understand and respond faster, smarter, and cheaper all can be enhanced using algorithms. However, without the appropriate level of customer understanding and strategic topspin, misplaced algorithms can simply leave the customer feeling alienated.

 

Specific purpose, specific result

With so much fanfare, it’s hard to blame the marketer who says, “I’ve got this awesome algorithm that removes decision making: there are no humans involved and I don’t have to think about it. I’m going to lunch!” But is that marketer genuinely looking at the process from the customer’s point of view? Was the messaging based on understanding the customer’s needs and did it address them?

 

Can the algorithm make a customer feel “hugged”?

 

Mostly, no.

 

As a customer, if you’ve ever been chased, maybe even startled, by ads and messages that seem to follow you wherever you go online, you probably agree. That’s stalking, not hugging.

 

Algorithms often identify customers who have a high likelihood of purchasing or attriting. Then, leveraging AdWords, search words, or numerous other advertising mediums, brands follow these customers at an aggressive pace. It’s a less-than-ideal approach. Simply identifying a quality target and then being in their face for the foreseeable future does not maximize customer value or take the whole picture into account. 

 

These efforts to be ever-present in front of potential shoppers will in many cases drive short-term lift. However, say we receive one million views on our ad and because of that we generate $20K in sales. Great. But how many people viewed the ads who didn’t buy? What was their potential future value? There’s a lost revenue impact that is rarely realized—losses that aren’t included as measure of success.

 

When algorithms repel

It’s about more than overlooking profit. An algorithm that retargets abandoned shopping carts and pounces on customers at every turn is flat out alienating. From the business function, this is great; the hard-working algorithm is proving its worth. But from the customer perspective it’s a different story.

 

After visiting a brand’s site, have you been stalked by its ads? I have! It’s creepy. Many people feel Big Brother is breathing down their necks on the web. Google and Facebook’s data algorithms have set off negative reactions from customers, as well as from senators. Uber surge pricing—using data to drive up the cost of ride sharing when there’s high demand—ticks people off. But the algorithm that sets the pricing is looking at the short-term gain versus the long-term loss.

 

There are unlimited marketing examples where algorithms make sense from the business perspective but fall short from the customer perspective.

 

The culprit, the algorithm, is simply doing the job we trained it to do. It’s producing short-term lift by identifying an audience to go after with no consideration of future consequences. Not all algorithms are created equal and many need strategic, customer-focused topspin to deliver the optimal results.

 

The benefits to automation are clear: If we can remove a human from the process, we can act faster while freeing up resources to do something else. We need to look at it from the customer experience, though. Ask yourself, “If I was the customer, would I enjoy this is?” If your response is “Heck, no!”, call a timeout.

 

Customers are getting smarter every day. Marketers who treat them as meandering sheep with open wallets full of money ripe for the taking will get trampled by the herd.

 

The fact is, all customers are looking to have a good experience. It’s important

to them. That’s why we request ratings, court customers to have them take a survey, and covet their positive feedback. All of which should work hand in hand with the algorithms.

 

To do it right, you need a human component.

 

Humanizing algorithms (i.e., being hug-worthy)

The algorithm is kind of like Willy Wonka’s Secret Machine. It does something, but without Mr. Wonka there to explain exactly what, the algorithm can easily be leveraged incorrectly and create confusion around the output.

 

At the end of the day, it’s about customer relationship marketing, and a relationship needs human engagement. Functions and process don’t build relationships. The process needs “humanization.”

 

It’s also critical that the data scientists and analysts who create these algorithms work closely with marketers to deploy them. Algorithms need to be built in context and deploying them should factor in elements such as lifecycle stage, the customer’s engagement with the brand, and what type of message will best resonate based on the algorithm’s predicted behavior. In other words, know when to send a “hug” versus a “push.”

 

When trying to identify customers who are likely to attrite, for example, an algorithm may deliver thousands or even millions of records that represent customers who are in significant danger of never returning. From the algorithm’s perspective, these customers simply become a 1 or a 0; they’re likely to attrite or they’re not. The default action on information like this might be to deliver a deep discount and a dramatic subject line and try to spark some renewed interest.

 

However, with a closer look from a strategic analyst and a lifecycle marketer, these customers represent very different behaviors and value to the brand. There are most likely some historically high-value customers who are trendsetters that buy only newly released products and never buy on discount. These customers may represent the highest historical and potential future value. Yet the action we’re taking to save them, providing a deep discount, can potentially do more harm

than good.

 

Algorithms provide immense value, especially as we collect more and more data about our customers that we can use to deliver targeted and relevant messaging. But these algorithms cannot operate in a vacuum. They need to be built into the human experience. We as marketers should be telling the algorithm what we need to know and letting it inform the process. Don’t let it drive the human experience.

 

A good algorithm enables activities and communications that make sense at the individual level to be done at scale. When you incorporate personalization and create the optimum customer-centric experience that is gauged according to human reception and desired results, then you have the formula for success.

 

Now get over here and give me a hug!

About the Author

Nick Godfrey is EVP and cofounder of Customer Portfolios, a marketing technology firm that uses insight and analytics to increase customer value.  Find Nick and Customer Portfolios at @CustPortfolios on Twitter.

 © 2019 MKTGinsight/DMCNY

 

  • Twitter Clean
  • White LinkedIn Icon