My local shop manager can anticipate my behavior in a way that makes me feel totally undisciplined every single time. I will get a WhatsApp message from him with pictures of several large Ben & Jerry’s ice cream tubs, accompanied by a cheerful message: “Sir! I just had a delivery. Shall I send one?”
My mind and my responses go through the same pattern each time:
“No, thank you, sir. I can’t this time. I should be losing weight.”
“But it’s been a really difficult week.”
“So, I do deserve a treat while watching the latest episode.”
“If I buy the whole tub today, I’ll just have a little bit and then keep the rest for the month.”
“Okay, I’ve changed my mind! Please send me a tub of New York Fudge Brownie.”
With the best intentions, I take just little spoonfuls from the tub while watching TV. I really don’t know how it happens, but somehow, it empties so quickly!
I compare this with the constant emails I get from Amazon telling me they’ve found something I might like. Invariably, it’s more of the same. There’s a viral meme that captures this perfectly: “Dear Amazon, I bought a toilet seat because I needed one. Necessity, not desire. I do not collect them. I am not a toilet seat addict.”
So why do they do it?
Failure to understand human utility: E-commerce systems are brilliant at recognizing that Product A is highly correlated with Product B. If you look at a toilet seat, the algorithm tags you with a “Bathroom Hardware” interest. It cannot distinguish between a consumable you might buy weekly, such as coffee, and a durable good you buy once a decade, such as a toilet seat or washing machine.
The data silo blind spot: Often, the marketing engine that follows you around the internet with ads is detached from your actual purchase history. It registers that you spent 20 minutes intensively researching and clicking on Bemis Round Toilet Seats or Church Elongated Seats, but it doesn’t “know” that you eventually completed the purchase. It just sees intense interest and assumes you’re still shopping.
The numbers game: From a purely mathematical perspective, ad space is incredibly cheap to run. Even if 99% of people are annoyed by being treated like toilet seat collectors, if just 1% of users are landlords, contractors or people who realize they’ve bought the wrong size, the automated ad pays for itself.
In other words, prediction does not necessarily tell you WHY the behavior occurs.
Yet we love this data-driven approach because it makes us feel we’re being rational.
We work in a culture biased toward things that can be measured. The underlying assumption is effectively that if we cannot measure something, it is less real.
There is pressure to sell more of what already works. In other words, businesses naturally favor exploiting known behavior to sell more units.
The culture doesn’t ask what customers are experiencing. Instead, we ask, “Can I defend this approach to my boss?”
Scariest of all, the customer becomes a mathematical object to be optimized. The language used in meetings is very revealing. Words such as target, conversion, propensity, funnel, acquisition, lifetime value and churn prevention are commonplace. I’m not suggesting that any of these terms is inherently wrong, but collectively, they encourage us to view customers as statistics to be incrementally pushed toward a desired outcome.
The question becomes, “How can we make them behave in a way that we want?” rather than, “That’s interesting. Why are they behaving this way in the first place?”
Organizations that are genuinely customer-behavior-led have cultures that are very different.
They have a high tolerance for not knowing. Senior people will say, “We don’t understand this behavior,” without losing credibility. Curiosity is rewarded more than certainty, and everyone is expected to challenge the category itself. Senior people ask more open-ended questions and listen much more.
Frontline employees are respected and listened to. This is the number one test. Organizations that patronize their sales staff with HQ-designed selling scripts are guaranteed to miss out on what’s really going on.
The language used in meetings also shifts from targeting to understanding. Words such as context, need, job, friction, motivation, trigger, barrier to change and workaround become more common. In other words, the customer is treated as a person operating within a subjective context, not as a data point.
If I were to compare the two approaches directly, I would say: prediction cultures seek internal confidence, while behavior-understanding cultures seek external explanations.
The former may sell more in the short term, but only the latter can innovate in the long term.
How do we combine the two?
Open up your favorite AI tool and ask it to explain “extreme users” and “extreme non-users” in design thinking. This is one way an organization can gain insights through a data-collection model while still examining behavioral outliers that may lead to insights and innovation across its entire customer base.
I presented these examples to Etihad’s senior leadership during an innovation training session. In each case, the individuals are using the same features of the Etihad lounge product but have different “Jobs to Be Done” (please see my previous article).
Alex, the Arrived Flyer: Someone who has traveled extensively for work has finally earned that Gold Card! How do they expect to be treated and recognized on their first lounge visit?
Maya, the Protective New Mother: Maya has to travel without her partner with a new baby. She’s anxious and doesn’t want the baby to disturb other flyers. She has never flown business class. What expectations might she have of the lounge staff?
Peter, the Gate Hawk: Peter is totally paranoid about missing his flight and always seems to find himself short on time. How can you use data and the flight board to reassure him?
All these “extremes” can be identified and served with empathy. This approach encourages organizations to move away from the idea of the “typical customer.” Yet the data collected can deliver new ways of serving the entire customer base.
Working with your favorite AI tool to identify extreme users in your business category is the first step toward understanding motivations and then using data to dig deeper.
It’s not surprising that my local shop understands my need for ice cream. Note, however, that it wasn’t Netflix that came up with the expression “Netflix and chill.” It was users who understood their own context after a hard day’s work. What other contexts might you discover in your category?
(Anand Vengurlekar is an award-winning senior lecturer in innovation and design thinking at INSEAD, Bayes Business School, IE/FT, PwC, Korn Ferry and other institutions. He focuses on mentoring individuals and teams to become more innovative under stressful circumstances and high-pressure deadlines.)
