Category: Halla Blog

Something is Rotten in the State of Online Grocery Recommendations

Posted on by Gabriel Nipote

You are grocery shopping online. One by one, you type in keywords for the items you need and compare the images, descriptions, and prices that are suddenly laid out before you in backlit splendor. It’s fast, it’s easy, and you don’t have to schlep a heavy cart and a couple of kids through the narrow aisles of a physical store.

But there’s something else happening as you fill up your virtual cart with culinary consumables. Below or to the side of the items in immediate view are recommendations of products the store has thoughtfully considered you might also need, based on your current selection.

And…they are all simply terrible.

Three ways that online food recommendations are undercooked.

Anyone who has done any grocery shopping online is probably familiar with the above scenario. Recommendations are responsible by many accounts for up to 35% of the transaction volume of those e-commerce players who are getting it right. It isn’t any wonder that they have come to be seen as a competitive factor in grocery as well.

But when it comes to online food shopping, recommendations tend to be pretty rotten. Here are a few common examples:

1. Recommendations that are too general

Bananas-768x1024.jpg

An open secret of grocery recommendations is that the algorithm seems to always think you might be looking for one item in particular—bananas. The reason behind that is confidence. No, not the kind that the grocer has in the freshness of his or her produce, the kind that a machine learning algorithm uses to quantify the probability that some “Event A” will coincide with another “Event B.”

Since the humble banana is consistently one of the top-selling grocery items, its purchase happens to coincide with, well— just about everything else. So no matter if a shopper has just selected basil and garlic or if it is spatulas and beer, over-confident algorithms trained on narrow input mechanisms like SKU and UPC data are going to proffer up the very same suggestion.

Imagine, instead of making generic recommendations based on broad purchase data, that the algorithm “knew food.” It could reach into its knowledge base of recipes and its awareness of “food psychology” and cleverly recommend pine nuts, olive oil, and fresh parmesan cheese to go with that basil and garlic. And then instead of offering you something quite irrelevant, presto…you’d have pesto!

2. Recommendations that are just plain wrong

Another bad recommendation scenario relates to a machine learning concept known as “lift.” Whereas confidence considers the general probability of A and B coinciding together, lift calculates the specific probability of “Event B” occurring when we already know that “Event A” is underway.

That may sound more promising, but it, too, can produce some strange results.

In that scenario, the grocer is not just losing out on the chance to increase their checkout value, they are actually alienating customers with annoying, or even insulting suggestions. That can make for a hairy situation indeed.

Imagine, for example, that amongst 100,000 shopping transactions, 100 include the purchase of a pack of Lender’s frozen bagels, and 50 result in the sale of a Panasonic nose hair trimmer. Neither quantity is enough to have any significant statistical impact.

But now imagine that, just by coincidence, 25 of those Panasonic buyers also happened to be fond of frozen bagels. The recommendation engine would “learn” that one out of four bagel purchasers has a pileous proboscis, and would begin suggesting Nova and a nose-trimmer to go with your Cinnamon-Raisin.

In that scenario, the grocer is not just losing out on the chance to increase their checkout value, they are actually alienating customers with annoying, or even insulting suggestions. That can make for a hairy situation indeed.

3. Recommendations that aren’t for you

There’s still one more mistake that general recommendation algorithms tend to make in the grocery world—failing to calibrate to personal tastes. When it comes to how we eat, variance is extremely high. So even when a recommendation is specific to a particular purchase journey and makes perfectly logical sense, it can still go wrong by failing to recognize a shopper’s personal preferences.

Bagel-687x1024.jpg

Let’s go back to our Cinnamon-Raisin bagel buyers. But now, the algorithm suggests Philadelphia cream cheese instead. Statistically and logically, we can’t fault that, right?

No, except that if our shopper, in this case, has a history of buying lactose-free dairy products or even organic, dairy and nut-free, vegan, paleo, fair-trade cream cheese substitutes, our recommendation engine would serve both grocer and shopper better by suggesting a product more appropriate for that particular shopper.

Likewise, while a bodybuilder with sliced turkey in his cart is likely to respond well to high-protein, low carbohydrate recommendations like yogurt, chicken breasts, and Brussels sprouts, a mother of three buying the very same item actually needs sliced bread, sandwich bags, and Capri Sun pouches for next week’s school lunches. And please Mom— hold those yucky Brussels sprouts!

Recommendation algorithms were never designed for food before

Considering all that is commonly wrong with online grocery shopping recommendations, we must ask—”What’s going on here? Why have online grocery merchants failed to emulate the success stories of companies like Amazon and Netflix?”

It is simply not possible for a recommendation algorithm to be broad enough to generate product recommendations for industries from beauty to electronics and simultaneously be specific enough to have a deep grasp of a complex industry like grocery.

The technology companies who lead innovation in the online recommendations space are full of extraordinary talent. Their product algorithms exhibit an uncanny understanding of human decision-making nature. But like almost everything in nature, there’s a trade-off—in this case, between breadth and specificity.

It is simply not possible for a recommendation algorithm to be broad enough to generate product recommendations for industries from beauty to electronics and simultaneously be specific enough to have a deep grasp of a complex industry like grocery.

Artificial intelligence that is actually intelligent should be based on highly relevant and industry-specific contextual data and abstract qualities, not just general purchase data.

For grocery recommendations, that means deploying an algorithm which has been trained on recipes, ingredient lists, nutrient disclosures, product attributes, and call-outs. It also means coding in a true understanding of what past purchases reveal about a shopper’s dietary habits and tastes, so that recommendations can be personalized.

One-size-fits-all recommendation engines based on simple shopping heuristics are a recipe for failure in online grocery shopping recommendations. To achieve the full potential of AI in online grocery, you need an algorithm built specifically for grocery.

Using anything else would be, well—just bananas.

What is “Food Psychology,” and Why is it Important to Online Grocers?

Posted on by Spencer Price

A typical large grocery has more than 30,000 SKUs, and yet the average grocery shopper purchases less than 300 of them over the course of any given year. Why is that? What determines the individual choices that food shoppers make? And how can grocers understand what their customers really want?

These questions and others speak to the heart of ‘food psychology’ —that set of hard-to-define values, attitudes, tastes, and behaviors that determine what a shopper is likely to pick up in a trip down the real world or virtual shopping aisle.

To begin, are your shoppers health-conscious buyers, who are willing to pay more for free-range, organic, non-GMO products? Are they family staple buyers, looking for the most nutritional value, cost and convenience for their shopping buck? Or are they epicurean, indulging in the best of gourmet options like imported wines and fine cheese? The beauty of online grocery shopping is that all of this can be determined from a shopper’s purchasing behavior.

Most online grocers are getting food psychology wrong

Where food psychology really becomes important in the online grocery world is when you want to offer highly relevant and helpful recommendations to users during their online shopping process. Acting on evidence that AI-powered recommendations increase retail purchase volumes up to 18%, it is understandable that grocers are incorporating this into their arsenal of e-commerce tools.

But more often than not, the recommendations online grocers make fall flat. Identifying shoppers based on the kind of simple lifestyle preferences stated above is too simple an approach. A true understanding of food psychology requires that we go much further.

For example, what if our presumptive epicurist was really just preparing for a fancy dinner party during their last session? That imported caviar she purchased was less an indicator of her genuine preferences than temporary circumstances. Likewise, although our family value buyer usually takes home staples like milk, eggs, and flour, that doesn’t help you understand that the ground beef, tomatoes, and onions in his cart actually reflect an intent to make a Bolognese sauce. If you knew that, you might have recommended dry white wine, garlic, bay leaf, and tagliatelle instead of another basic staple like bananas.

And again, perhaps our presumptive health-conscious buyer sometimes shops specifically for a vegan non-drinker, but also for a partner who prefers their steak bloody and their liquor strong. Recommending only Tofurkey patties and artisanal kombucha may not get you as far as you’d hoped.

When grocery shopping recommendations are made on narrow data and simplified paradigms without really understanding shoppers’ complex relationship with food, grocers face double-jeopardy: Not only do you miss the opportunity to earn a bigger shopping cart checkout, but the poor recommendations also annoy and alienate the shopper, who feels “sold to” instead of “served.”

There is a better way, and it’s really pretty intuitive.

To give good online shopping recommendations, online grocers need to:

  1. Take an inventory deep dive
    Grocery items are remarkably data-rich. Consider a typical grocery SKU: The name, manufacturer’s description, price, category, and brand reveal whether this is a food product or a non-food product, and whether it is something applied, such as lip balm, or something to be used externally, like a spatula. Most food products can be further broken down into myriad sub-components according to their ingredient list, nutrient disclosure, and product attribute call-outs such as “free-range” or “microwave” that reveal important psychological attributes of consumers’ food preferences.

    The way to then make actionable sense of all of that product information is through artificial intelligence techniques like natural language processing. NLP gives grocers a structured way to convey sophisticated human knowledge about food SKUs to a computer algorithm: “What is a protein?” “What is a vegetable?” “What is a grain?” “What is an alcoholic beverage?” and “What is a condiment?”—but also, “What is healthy?” “What is convenient?” “What is good value for money?” and “What is indulgent?”
  1. Every customer is unique
    As we discussed above, while it may be useful to think of customers in terms of well-defined avatars, the reality is that each shopper is likely to express a wide range of purchase behavior based on what they are preparing, who they are shopping for, and variations in their personal preferences over time. It is never enough to simply pigeonhole a customer into a simplified caricature of shopper type. Instead, you must put together whatever you know about the shopper’s personal characteristics with past purchase behavior, response to previous recommendations, inferences from products purchased, and a wide range of other clues each customer displays during their interactions with your platform. Your recommendation engine must be built to serve market segments of “one,” and the work of defining who they are is never done.

  2. Buying versus Shopping
    Understanding products and customers is an important start. But to really present the right recommendations at the right time, you must also understand the actual context of the purchase. At the simplest level, that can refer to seasonal variations (imagine receiving “pumpkin spice” recommendations in the middle of July) or geography (preference in sweeteners, for instance, vary dramatically amongst different regions and populations).

    But understanding the contextual relationship between the items a buyer has already put in their cart can also make all the difference in the world between good and bad recommendations. There is a unique relationship, for instance, between flour, butter, and a dozen Granny Smith apples—picking up that these belong to ‘from scratch’ apple pie recipes empowers you to propose complementary items like cinnamon, nutmeg and pie pans that customers really need.
  1. You are what you eat (so what are your customers eating?)
    And really that brings us to the most complex and fascinating requisite of making good grocery recommendations— understanding food itself. Each one of us has our own personal flavor preferences and dietary habits. To get to the heart of what online grocery shoppers want, your recommendation engine has to know food, inside and out. That means gathering and classifying, millions of food ingredients, flavor profiles, products, dishes and recipes, and then building a machine learning algorithm with an unparalleled level of understanding of how customers relate to those attributes.

    Grocery recommendations have a unique logic which can not be served by copy-pasting the kinds of recommendation algorithms used successfully by entertainment and general e-commerce players. To make really great grocery recommendations you need to break down your products into recognizable data, understand how your customers shop, take into account contextual variation in shopping habits, and then build an AI model that is specific to the psychology of food. Machine learning offers the ability to do this at a scale that humans could not dream of alone, and Halla is proud to offer the only truly grocery-specific recommendation engine on the market today.