Algorithms: The New Tastemakers
JULIA SEGAL & HALEY KABUS,
Sr. Strategists
Let’s pretend that I’m an algorithm. It’s my job to whittle down a seemingly endless stream of possibilities into a manageable spectrum, giving you - the user - what I know you love to see. I ensure that you and others who consider themselves foodies of the more esoteric variety have a collective understanding of what makes an egg “jammy” or what makes a wine “glou-glou” instead of quaffable or why offset spatulas are essential for crumb-coating multi-layered cakes. I’m also responsible for helping you join the ever-growing mass of epicureans simultaneously baking Samin’s focaccia or #TheCookies from Alison Roman. I’m quite good at what I do, so much so that you barely notice me. I give you what I know you want, what’s the harm in that?
In childhood, discovering new foods, dishes and flavors is as simple as taking a bite out of what’s put in front of us. Over time, our palates develop into our food preferences. As we further grow into ourselves, eating and exploring new foods can be identity-building. Who am I, Julia Segal, without my magnificently stocked bar? Or I, Haley Kabus, without my rotating display of opulent desserts? Eventually, we learn just how niche or adventurous we are as consumers.
Nowadays, we don’t need to leave our homes, or even our couches, to discover new flavors. But in our search for food novelty — digital or real world — there are algorithms that dictate what we see and what is omitted from our feed. As we aim to explore, are we truly broadening our options, or are we actually just being fed subtly different versions of the same, further solidifying our niche preferences? As algorithms aim to push the content we love most, aren’t they also perpetuating a confirmation bias that will lead us deeper into preference-reinforcing rabbit holes? With the intentional suppression of the ‘unfamiliar’, how will regain access to truly novel food media and experiences?
Confirmation bias and the narrowing of interests.
Everyday we interact with thousands of algorithms. When we listen to Spotify or Pandora, scroll through Instagram or TikTok, watch Netflix or YouTube, we are feeding, interacting and benefiting from algorithms. While there is a lot of complexity in this space (that we won’t pretend to fully understand), there are some known underlying principles.
For example, these content and media algorithms share a common goal: increasing time spent watching, playing and interacting. Time spent is both an indicator of satisfaction and a path to more advertising sales. So what’s the key to these algorithms keeping people engaged? Showing us what they know we already like, based on our past activities, our demographics, our location (basically any information they can get their hands on). Chances are, the vast majority of what we are exposed to reinforces and expands on our existing preferences.
While these algorithms are not inherently evil, in 2020, as more people spent much more time online, the dangers of these suped up recommendation engines became much more apparent. Falling down a rabbit hole of a myopic and niche world view is easier and faster than ever.
When we turn to food media, we see traditional tastemaking endeavors being replaced by the power of going viral on one of these tech platforms. In March 2020, everyone was making sourdough. In July, everyone was whipping up their own dalgona coffee. In December, everyone was making hot chocolate bombs. None of these quarantine trends started on the pages on Bon Appetit or in a restaurant, they began on my feed. #dalgonacoffee has been viewed 440.5M times on TikTok, making it the Superbowl of TikTok food trends. But for every dalgona, there are thousands of culinary niches for people to fall into. Like @bashan0915, who has 200k followers of his serene, ASMR outdoor cooking in China. Or @thekatcurtis, who combines two weird foods - like sour worms and hot dogs, ramen and cheetos, peanut butter and pickles - amassing 1.7m followers in the process.
Whether building our digital food & beverage personas on TikTok, Pinterest or Instagram, our data is a tool for restaurants to use to recommend items that they already know are likely interested in. But as exposure to food & beverage content as part of our highly curated feeds, turns into exposure to food & beverage items IRL, how will our palates evolve?
The burgeoning world of restaurant A.I.
In 2020, the majority of our restaurant experiences and interactions moved online, forcing many businesses to focus on adaptation, rather than innovation. Now that the importance of the digital ecosystem to our industry has been cemented (for both virtual and brick ‘n mortar brands) we believe that the pace of innovation will increase rapidly.
One of the most intriguing places of food tech innovation has collided with the most valuable pandemic real estate - the drive thru. In 2019, we saw McDonald’s bring AI to drive-thru. And today, McDonald’s investment in AI and machine learning has breached $300M, a signal that big food is becoming more like big tech, complete with innovation labs in Silicon Valley. And so far, the goals for restaurant AI seem similar to those we’ve seen in retail companies, social media platforms, and content providers - to increase the amount of curation and prediction around what their customers want.
At McDonald’s drive-thru, if you opt-in, a license plate reader will log your order and use the data to tailor your future recommendations. If you order a Big Mac today, they think you’ll want one next time, and that suggesting it to you before you even order will make you happy.
But, how does restaurant AI that suggests what we want, based on what we’ve chosen before, change our eating in the long run? If big tech has taught us anything in 2020, it’s that indulging our confirmation bias, our existing tastes, can alienate, divide and disillusion us. Regardless of what’s happened, is restaurant AI that recommends what we’ve had before the best use of this tool? As more of our restaurant interactions turn digital, there’s a more valuable role for AI to add to the experience - a human element.
The future of restaurant A.I. is harnessing discovery and old school hospitality.
When ordering food online, we have come to expect the app to recommend an upsell - a side of fries or a cookie. In traditional dining, the same is true, though in a more subtle way. Restaurant servers are masterful at casually mentioning an ideal side dish, a perfectly paired beverage or a not-too-sweet finisher to a meal. In a restaurant setting, we chalk it up to hospitality - a good server anticipating our needs. In the app, it’s a blatant attempt to increase the check average.
In order for restaurant AI to carve out a meaningful space within hospitality, it needs to hone in on closing the gap between predicting your order based on previous preferences, and anticipating needs, wants, and moods. Anyone who's been a regular at a coffee shop knows the power of your ordering being ready without you having to ask. Despite the sophistication of AI within our ordering platforms, it hasn’t cracked the code on balancing discovery and familiarity. Short of asking us outright, the next step is for the algorithm to assess if today is for exploring new flavors or sticking with the tried-and-true.
There's an opportunity to understand a guest’s tastes and preferences beyond the item level. Being able to tap into a more targeted understanding of a guest’s preferences and make ingredient-level recommendations is a powerful tool for building loyalty. Not only suggesting a particular sandwich, but knowing to swap goat cheese for cheddar, and add a scoop of hummus relies on a deeper level of understanding our consumer’s needs.
Building on this, how can we anticipate future contexts, like forecasted consumption trends, to provide suggestions for discovery. In a traditional hospitality setting, interpersonal cues, environmental factors and having a finger on the pulse of consumption trends play into how we best serve our guests. AI has cracked many codes, but figuring out how to emulate the “best server ever” is the next frontier.
In order to remain competitive as we digitally connect with our guests, we need to move far beyond offering them what they like, adding a side of fries, and promoting LTOS. We need to lean into the hospitality element that makes our relationship with restaurants so unique. We need to use as much future-focused forecasting data as we do consumer behavior data.
What is the version of the digital platform that knows you need a comforting bowl of soup because it’s cold outside? When do we suggest a nicer bottle of wine because this week was an important milestone?
Restaurants will never truly win in this space by asking if their guest wants fries with that? We need to move the goalposts to where upselling isn’t the ultimate measure of success, and providing an experience of incomparable hospitality is.