Amazon Fresh Series - Part 2: Customer Decision Making
Readers who searched for this also read this:
Everyone who has used some form of consumer technology has read or encountered some version of the headline above. Technology companies use headlines like this to help influence our next behavior. Will we finally find the camera we were looking for? Will we read through the entirety of the next article? Will we end up buying 4 items instead of the 1 that we actually logged onto to the internet to buy? In physical retail, companies have used item and category placement to move customer unconsciously around the store in the hopes that the answer to the last question is a yes and the last minute milk you needed for after dinner coffee and pie will be accompanied with 2 pints of Ben & Jerry’s and Cool Whip (What I would call a “Need to Have”). Milk is a great example of this strategy as most grocery stores put the milk in the farthest corner of the store from the entrance in order to make sure the customers have to walk as much of the store and pass as many items as possible in order increase the impulse purchases of your trip. Your son, who has reluctantly joined you to the store, has been glued to his phone the entirety of the trip and you see him scrolling through his Twitter Feed. What you didn’t expect was that when your trip to the grocery has been replaced with a visit to your online grocer of choice the same social science is being applied to your e-commerce purchase that your son is also experiencing as Twitter tries to place the most relevant articles and posts in front of him.
Introduction to Captology
Using technology to as a means of persuasion to change a user’s behavior was coined Captology (Computers as Persuasive Technology) by BJ Fogg a Stanford Graduate student in the mid 90s studying human psychology. Fogg went on to found the Persuasive Technology Lab at Stanford a few years later. Now it is known as the Behavior Design Lab in order to remove the bias and better incorporate the ethics first approach that the academic institution has taken for this work. The teachings have helped to blossom some of the biggest consumer technology platforms of the last 2 decades and its Alumni have gone on to found and accelerate the growth of these companies worth hundreds of billions of dollars including Facebook and Instagram. This Economist piece, published in their 1843 Magazine, does a really great job outlining the topic through a Behavior Design Workshop that Fogg leads as well as talking about why these same technologists are worried about the theories misuses. The article outlines Fogg’s overview of how to change someone’s behavior:
“For somebody to do something – whether it’s buying a car, checking an email, or doing 20 press-ups – three things must happen at once. The person must want to do it, they must be able to, and they must be prompted to do it. A trigger – the prompt for the action – is effective only when the person is highly motivated, or the task is very easy. If the task is hard, people end up frustrated; if they’re not motivated, they get annoyed.”
“The trigger, if it is well designed (or “hot”), finds you at exactly the moment you are most eager to take the action. The most important nine words in behaviour design, says Fogg, are, “Put hot triggers in the path of motivated people.”
Ok, so you are probably asking at this point in the article why on Earth I am going into so much detail about this? The theory introduced in this study has helped to produce and describe some of the most important pieces of describing how Amazon has become such a powerhouse in the e-commerce market.
1-Click Buy
One of Amazon’s most important “hot triggers” is the invention of the 1-Click Buy button. The technology behind this button was patented by Amazon in 1997 and has recently expired in 2017, effectively giving them a 20 year head start. They aggressively defended this technology, suing Barnes & Noble in 1999 and licensing the tech to a few companies including Apple’s iTunes. R. Polk Wagner a professor at University of Pennsylvania’s Law School best describes the two most important reason why this technology transformed Amazon:
“Most importantly, it allowed Amazon to show customers that there was a good reason to give them their data and the permission to charge them on an incremental basis. It opened up other avenues for Amazon in e-commerce. That is the real legacy of the 1-Click patent.”
Not only did it pave the way for Amazon Prime memberships as Wagner alludes, but it created one of the best “hot triggers” in e-commerce at the time. In physical retail, a common place where Market Researchers reference as creating friction in the Customer Journey is at the checkout. However, with e-commerce, the problem at the time was arguably worse than at the physical store. Each time you wanted to purchase something you had to type in your mailing address, your billing address and then find your wallet in order to put in your credit card information. Compare entering your information into the e-checkout to the 18 items that Aldi expects its cashiers to checkout per minute and you immediately see where the problem really was in retail. (It’s important to note that even though the patent expired in 2017 many retailers still do not have the option for this on their website!) Ron Berman of the Wharton School of Business also at the University of Pennsylvania effectively outlines that the e-commerce checkout issue was exacerbated on the mobile device.
“Because [mobile] screens are small, the larger the hassle [or number of clicks] it is to purchase, the lower the purchase propensity on mobile phones…”
This capability significantly lowered the friction for online shoppers over the next two decades and was an undeniably important part of how they were able to convince shoppers to start their searches on Amazon and one of the main reasons why 63% of the e-commerce searches start on Amazon.
Last but not certainly least is the impact that 1-Click Purchase Button had on e-commerce cart abandonment can not be understated. According to the Baymard Institute, an Independent Web UX Design Research Institute, over the last 11 years 69% of online carts are abandoned. They have found that the main reason for cart abandonment was that 58.6% of shoppers described their activity as “I was just browsing/(am) not ready to buy” (which we will dive into in the next sub-section of this article) however when they segmented these answers out the following reasons were outlined:
It is hard not to see how Amazon has solved the 3 next main reasons for cart abandonment with Amazon Prime and the 1-Click Purchase Button. To illustrate the size of the prize, the Baymard Institute dove into the third reason - “Too Long/complicated checkout process”. Based on their research the Average US e-commerce checkout has 23 form elements where the optimal is around 12-14 elements and by optimizing these same elements, a large sized e-commerce business can gain a 35% increase in conversion rate - or about $260 billion worth of lost orders in the US and EU last year. That is just less than all of Kroger, Costco and Albertson’s Conusmable sales for US and Canada in 2019 at $254 Billion.
Amazon Recommends you read this next section
Amazon hosts a Science Blog where they introduce and outline the research that their team are and have completed across all parts of their business. Last November they published an article about the history of their Recommendation Algorithm. The article not so subtly starts by describing that the Academic Journal IEEE Internet Computing named Amazon’s 2003 Research “Amazon.com Recommendations: Item-to-Item Collaborative Filtering”, by then Amazon researchers Greg Linden, Brent Smith, and Jeremy York, as the single article from their 20 years in existence to most clearly stand the test of time. Describing the paper and it’s use cases Amazon writes:
“Collaborative filtering is the most common way to do product recommendation online. It’s “collaborative” because it predicts a given customer’s preferences on the basis of other customers’.
“There was already a lot of interest and work in it,” says Smith, now the leader of Amazon’s Weblab, which does A/B testing (structured testing of variant offerings) at scale to enable data-driven business decisions. “The world was focused on user-based collaborative filtering. A user comes to the website: What other users are like them? We sort of turned it on its head and found a different way of doing it that had a lot better scaling and quality characteristics for online recommendations.”
The better way was to base product recommendations not on similarities between customers but on correlations between products. With user-based collaborative filtering, a visitor to Amazon.com would be matched with other customers who had similar purchase histories, and those purchase histories would suggest recommendations for the visitor.
With item-to-item collaborative filtering, on the other hand, the recommendation algorithm would review the visitor’s recent purchase history and, for each purchase, pull up a list of related items. Items that showed up repeatedly across all the lists were candidates for recommendation to the visitor. But those candidates were given greater or lesser weight depending on how related they were to the visitor's prior purchases.
That notion of relatedness is still derived from customers’ purchase histories: item B is related to item A if customers who buy A are unusually likely to buy B as well. But Amazon’s Personalization team found, empirically, that analyzing purchase histories at the item level yielded better recommendations than analyzing them at the customer level.”
Over the last 17 years Amazon has not only dedicated significantly more math talent to developing these algorithms but has flooded the customer journey with these recommendations utilizing different angles in order to attempt to turn a browser into a buyer. These investments are not going unnoticed by the customer. A 2019 Feedvisor Report of the Amazon Consumer show that nearly half of the consumers notice these suggestions on the product page.
Illustrating how important these recommendations are to Amazon’s bottom line, McKinsey estimated in 2013 that more than 35% of Amazon’s sales revenue is generated from a recommendation that they put forward. Deploying these recommendations in both emails, third party website ads and most familiar to an Amazon shopper would be the recommendations on a products page. Looking across items on the Amazon Fresh page there are four main categories of recommendations that are being utilized. A look under the algorithm hood would help us better decompose each of these sections, but it’s more relevant to see these through the customer’s lens
1. Customers who viewed this item also viewed
Gus has clicked on a Happy Belly Pasta item and sees this section. This could help him find the right pasta item he was actually looking for or allow him to look at the price or reviews of these other items. Notice here that these recommendations are not suggesting any branded items on the first page of recommendations for this item. While they could be on pages 2-4, Amazon could have tailored the recommendations at this time in order to keep Gus’s focus on Happy Belly.
2. Customers who bought this item also bought
The next section Gus sees for the Happy Belly Pasta item is what other customers have purchased with this item. There are both “supporting items” like parmesan cheese or Ragu Pasta Sauce and also substitutes/additional pasta items that were purchased alongside this item. They key to this section is that customer’s have decided to buy these items together which is a traditional tool that retailers have used to promote cross shopping within their store and build bigger baskets. Nothing terribly new here.
3. Customers also shopped for
Gus is not just buying Happy Belly Pasta on this e-commerce trip. Amazon puts forward other items that Customers have looked for during their shopping trip that has the same Happy Belly Pasta that he is currently viewing. This can help customers find or build out their basket by either shorten navigation to other items that were on their shopping list or can help then explore/browse for new items to their current trip.
4. What other items do customers buy after viewing this item?
Gus is looking at the tortilla chips that he normally buys in his local store and sees the recommendation below that other customers buy Santitas Yellow Corn Tortilla Chips after they view the current item. He says: “ Well if other customers bought this item maybe I should try it too?”
Each one of these recommendation lists help to bring clarity to the shopper in order for them to solve for what product they ultimately put into their cart by addressing different shopper need states. However, as some readers may point out, these are not the only recommendations that Amazon uses across their marketplace. An additional recommendation that Amazon uses on it’s website, and is often the first one that is shown as the consumer scrolls down the product page is the Frequently bought together section. These combinations provide a singular button to add all three items to the consumers cart from a singular product page, in this case it is Hershey’s 6 count Chocolate Bars.
For your campfire s’mores, Amazon has suggested a branded marshmallow, (Jet-Puffed) and it’s own brand Graham Crackers!
This is incredibly powerful way to increase the visibility of cross selling in the shopper’s journey and is why ice cream cones are often positioned at the end of the freezer in your local store. However, in the physical store you will not find ice cream in-aisle next to the ice cream cones. Amazon can and does provide these cross selling opportunities in both or all of the items listed in the combination increasing the likelihood that your cart grows no matter what item you search for first. Thinking ahead for a moment - this strategy could be expanded quite significantly in order to further guide the customer on their shopping journey. The above combination is a recipe that we all know for S’mores, but given Amazon’s ability to scale technologies we can expect more complex combinations to be available for full meals in the not so distant future. This financial opportunity is clearly illustrated by a recent Zagat Survey shows that households, even post-COVID, will be cooking and preparing more meals at home showing that almost 60% of households plan to cook 5+ dinners at home, a 40% increase from Pre-COVID.
Retailers have worked to limit the barrier for inspiration into the consumer’s meal planning by investing heavily in prepared meals but cross-selling in the physical store, as mentioned above, can be item prohibitive as well as operationally intensive as these sets need to be assembled by in-store workers. Because of this consumers regularly do not find retailers as being a source of inspiration for their cooking adventures. FMI’s 2020 Grocery Shopping Trend Report shows that consumer’s family and Cookbooks lead the way in influencing with Grocery and other food stores lagging behind with other 13% of consumers stating that they provide the most inspiration for cooking.
Amazon could find the best of both worlds by partnering with Cookbooks and Chefs to create menus or meal suggestions using this recommendation engine in order to help consumers find the perfect meal for their friends and family and make it easier purchase these items. Taking it a step further, Amazon’s purchase history could help shoppers supplement what they have purchased in recent baskets in order to use items that they may already have in their pantry to avoid a common menu based shopping integration problem of purchasing supplemental ingredients like spices multiple times leaving them with 5 containers of cumin in their cabinet.
Lastly, the shopper will notice one other recommendation section in the Amazon Fresh product pages, Customer favorites.
Customer favorites are a combination of highly rated items that have a significant number of reviews behind the overall rating. As we will explore in the next subsection, Amazon’s rating advantage will be a powerful data point to help the consumers decide on in addition to exploring other products in the assortment.
***** Stars: Helping Consumers make more informed decisions
In 1995 Amazon ventured into what a lot of critics would be “retail suicide” by letting customers leave reviews of products in their store. Now reviewing what other customers say about products before you purchase them has become an important part of the customer’s online shopper journey. In fact, research released just last month (September 2020) by the Consumer Insights Team at Google (Highly recommend reading the full 98 page report published by the team) that explored the customer purchase decisions process, analyzed 6 of the most common cognitive biases associated with purchase decision outcomes:
Category Heuristics: Short descriptions of key product specifications can simplify purchase decisions.
Power of Now: The longer you have to wait for a product, the weaker the proposition becomes.
Social Proof: Recommendations and reviews from others can be very persuasive.
Scarcity Bias: As stock or availability of a product decreases, the more desirable it becomes.
Authority Bias: Being swayed by an expert or trusted source.
Power of Free: A free gift with a purchase, even if unrelated, can be a powerful motivator.
Reading product reviews has become ingrained in the Amazon shopping experience as Feedvisor shows that only 6% say they “Never” or “Rarely” read the Product Reviews before making a purchase.
As mentioned in the opening to this series Amazon has even opened stores that only displays products that are 4 Stars or above to help ensure trust with the consumer that the products they see are of high quality, hopefully eliminating buyer’s remorse. However, Amazon Fresh has introduced the Grocery Industry to a new practice that most food consumers will not have been exposed to before. When watching the Amazon Fresh Introductory Video there is a scene when a consumer is reaching for Happy Belly Flour and you can clearly see the price tag of the item. Other than the regular information, price, item description, size, you can also make out a series of stars with what looks like a small number in parenthesis!
Utilizing the user generated content of product reviews in the store helps the customer in a couple of really important ways and differentiates Amazon from the rest of the physical grocery world. Sticking with the product above, Private Label items are regularly pinned as lower quality than the National Brand equivalent and buyers see them as “cheap alternatives.” However, with the customer reviews on the price tag, the consumer can see that they will not only be saving 40% by purchasing the Private Label item, but that customers have actually rated it more highly than the National Brand Equivalent. This will certainly help Amazon accelerate their Private Label sales and household penetration within the physical grocery world to levels not seen by other retailers over multiple years of Private Label development and marketing. Emerging and Smaller Brands will also be able to equally leverage the power of the review on the shelf and should be allocating resources to Customer Service and Product Development to ensure that they “live” by the review. And most importantly, this adds a crucial weapon to the children’s “begging arsenal” to convince their shopping parent(s) to buy them the candy bar that they eyed going down the aisle.
“But it has 4.9 stars mom!”
Kidding aside, putting these reviews on the price tags has the potential to be more important that anything else that is talked about in this series and democratizes should be watched very closely by Brands, Retailers and Researchers.
To conclude, Amazon is way ahead on these categories and provide a significant initial moat to not only online grocery but in their new stores. Other Retailers and Brands need to focus on collaborating to create well researched and data first decisions to reduce the size of the lead that Amazon has created in reducing a significant amount of the friction for their Customer’s Decision Making.
Up next…the final installment, Assortment!
