You’re bound to walk away from the historic Yale campus with a notebook full of new ideas after being surrounded by leaders in insights and marketing. There’s no way around it—listening to the big-name speakers and chatting with the equally brilliant attendees, ideas start to fly as you consider how to implement the strategies you’re actively learning. With themes ranging from Elevating the Role of Consumer Insights to Strategic Marketing: Perspectives from Leading CMOs, there wasn’t a part of the conference worth skipping.
Ideas To Go has sponsored the Yale Customer Insights Conference for the past 9 years, and each year brings new and unique perspectives. As the years advance, so do the strategies of leading companies. To stay at the forefront of their industry, marketing and insights teams are constantly evolving their techniques, looking for better ways to reach their consumers.
In case you were unable to attend this year’s conference, we’ve got a 3-part series summarizing the main takeaways from each speaker and discussion. Here is part 1 of 3.
Will Platt-Higgins: VP of Global Client Partnerships, Facebook
If there is one phrase to sum up Will’s talk, it is Learn, Unlearn, Relearn—with an emphasis on UNLEARN. This follows the theme mentioned above—companies are constantly trying new strategies to better reach their consumers. And if you get stuck in your old habits, the experimental companies are going to pass you. Therefore, you need to unlearn what you think you know and explore new opportunities. To do this, Will highlighted 5 points.
- Your competitor may not look like you
What it takes to be a company today is not the same as what it took to be a company back in the day. Now, anyone with a phone can be a competitor. They have access to social media, they have the room for growth, and most importantly, they will be trying new things to captivate a listening audience. While you may be watching your competitors in the traditional sense, it’s the quick-risers that end up being the real displacers.
- Explore outside your orbit
50% of teens spend 5 hours a day navigating various social apps on their phones. That’s a lot of time for marketers to capture that audience—but some of those apps fly under the radar until they’re not as mainstream (see TikTok and Houseparty). Explore areas outside of the familiar to capture a larger audience.
- Mobile storytelling and :30
TV has been around since the mid-1900s, and while television advertising is still a great strategy, there is only one true format for the platform—video. Compare that to mobile advertising where content can be interactive and/or immersive. Virtual reality, images, videos, and augmented reality—where an ad exists on the phone user’s camera—are all ways to engage with your consumer outside of the traditional format. Embrace the complexity in format variance and experiment.
- We are shoppers
Gone are the days of in-store shoppers and gone are the days of mobile shoppers. Nowadays, we are just shoppers. Ready to shop at the click of a button, there’s never a moment we can’t purchase something. Shopping on a mobile phone combines window shopping and purchasing, but the physical store isn’t dead yet. Combine the digital and physical experience in a way that unlocks a unique experience for the shopper.
- The message from messaging
Messaging is the new customer service—customers don’t want to dial a number and work through an automated system to talk to someone. Messaging is their preferred method of contact, and marketers should embrace that. Consumers can now purchase via message, messages can be hyper-targeted through message, and the retargeting capabilities of messaging are endless. In fact, companies are switching from call centers to messaging centers, where live representatives take messages on iPads and respond to customer queries.
Dan Goldstein: Principal Researcher, Microsoft Research
As much as we like to perpetuate the fear of artificial intelligence, AI is everywhere in our lives. The first thing we do when we wake up is reach for our phones and unlock them with Touch ID or Face ID—a process driven by AI. Your email is constantly filtering out unwanted spam messages—a process driven by AI. Netflix recommends the next movie or series based on your preferences—a process driven by AI. Uber tells you how long it will be until your driver arrives—a process driven by AI.
Dan’s talk isn’t about the dangers of AI, but instead emphasizes the fact that AI should be interpretable. It is necessary to understand how AI is making the decisions it is making, and not just allow deep neural networks the freedom to do as they wish. Because if we don’t understand how our AI is making its decisions, how can we know it’s making the right decisions? This is especially worrisome in situations where bias can come into play.
The study Dan and his team conducted looked at determining factors of whether or not people should go to jail during their pre-trial interval. They wanted to see if they could set up an interpretable AI system to accurately predict whether someone would show up to court. The key to developing this AI is finding 3-5 factors that are correlated to past instances of people failing to show up to court, then assigning a simple point system to each. After each number is added up, there is either a yes or no answer for whether or not the person should be held in court until their trial. This sounds like a trivial system, but it is still a form of AI—most importantly, an interpretable AI.
The interpretable AI’s accuracy was then compared to human judges' decisions and a black box machine learning AI. The results? It outperformed the human judges’ decisions, and did just as well as the machine learning AI. And most importantly, we understand exactly why it made the decision it made. As it turns out, point systems with 5 predictors typically work just as well as complex AI models.
The catch is that there has to be enough correlated data on these predictors so that an accurate regression analysis can be done. Only if there are strong enough correlations can we then assign it to the point system. You won’t find image-recognition AI using a point system to recognize images of potatoes, because there aren’t any simple rules to set up for it to follow.
Complex tasks like image recognition still require complex AI, but interpretable AI should be used where applicable so that when a computer makes a consequential decision, we know exactly why that decision is being made.
Perspectives from the World of Insights: A Panel Discussion
Laurence Bucher - Global VP of Consumer & Market Insights at Mars-Wrigley Confectionery;
Stan Sthanunathan - EVP of Consumer & Market Insights at Unilever;
Ewa Witkowska - VP of Insights & Analytics at PepsiCo;
Moderated by Beth Storz - President and Innovation Process Facilitator at Ideas To Go
The first of two panels, our very own Beth Storz moderated this discussion on the world of insights. And as it turns out, the worlds of insights and marketing are beginning to blur. Their overlap is becoming more obvious due to the overload of data available on consumers. With this plethora of information, it is essential to identify what is important to you as a marketer, what is important to the consumer, and then invest your resources there. The difficulty is that the target is constantly moving. Consumer preferences change, then your insights change, then your target changes with it. It is the role of insights to identify these changes as soon as possible—sometimes even before they happen.
The panel emphasized the importance of investing in internal talent to achieve insight success. The people within your company are the ones developing the strategies, so you need to create a culture where people feel valued and want to perform their best. But it doesn’t end with finding the right talent—retraining programs need to be initiated to transfer pertinent knowledge to the entire program. Especially in today’s climate where job functions are constantly evolving, smart and adaptable internal talent is key.
Discussion then transitioned into being nimbler as an entire organization. As it turns out, PepsiCo is taking all validation in house so that their data is available to anyone within the company. The idea is to spread the information to anyone who needs it, instead of siloing it among departments.
One of my favorite moments came from Stan Sthanunathan at Unilever, who said they have started using a system called Concept Swipe, which he described as the Tinder for concepts. In essence, a consumer either likes or dislikes a concept, so why mess around with assigning numbers. Let the consumer swipe left or swipe right. The transition is similar to the one Netflix made, when they switched from their star ratings on movies to a simple thumbs up or thumbs down. Per The Verge, “Switching to a binary thumbs-up / thumbs-down system might seem less granular than offering five stars, but Yellin said there’s an implicit understanding with thumbs-up / thumbs-down that people are doing it to improve their own experience rather than trying to rate it for the rest of the world.” At the end of the day, you want to know if a consumer likes a concept—and the Concept Swipe answers that question.
Check back soon for Part 2 of the 2019 Yale Customer Insights Recap!
Want to attend the Yale Customer Insights Conference next year? Subscribe to Ideas To Go (see top of this post) for exclusive discounts on conference tickets. Interested in what went down at the conference in 2018? Check it out here.
Tyler Thompson is a Marketing and Research Analyst at Ideas To Go, an innovation agency that works with Fortune 500 companies in ideation and concept development to incorporate the voice of the consumer.