Week_11_Recommender_Fox
Intro: Walmart Online Grocery Shopping
Walmart.com has been a driving force in Walmart’s growth for the past decade. Between 2016-19, Walmart’s total e-commerce division grew a total of 78%, a rate twice that of Amazon’s 1. Their investment paid off even more during the pandemic: in 2021 alone, they saw a further 79% increase2. Groceries are a key part of their online strategy: in 2023, groceries made up 58.8% of sales3 and Walmart is now the largest online grocery retailer in the country4.
Walmart’s focus on online growth makes sense: with 90% of the US already within ten miles of a Walmart store3, online is the place to grow, especially as younger digital-native generations come of age. From a strategic perspective, Walmart has a significant competitive advantage over Amazon by blending e-commerce with its mature brick and mortar infrastructure: its stores provide convenient pickup and return options. Additionally, online shopping data and geoanalysis bolsters Walmart’s ability to continuously optimize the limited shelf space in each of its stores and warehouses.
In this paper, I review the design of Walmart.com’s grocery UX with a focus on recommender systems.
Walmart.com: Scenario Design Analysis
Who are the target customers?
According to a start.io blog article from 2021, a stereotypical Walmart shopper is lower-income, rural, and loyal to their local one-stop-shop Walmart store, while the typical Walmart shopper is actually a late-middle-aged suburban woman with $80K income, who shops primarily for groceries2.
Walmart’s demographic is likely to shift due to the rise of online shopping and younger generations coming of age: Walmart.com customers are demographically different from in-store shoppers (see below for income comparison), and price-conscious and tech-savvy Gen Z already names Walmart among favorite brands.
https://marketplace.walmart.com/9-things-about-walmart-customers/. Accessed 4/3/2024. Specific “typical” demographics aside, given Walmart’s incredible reach through stores, home delivery, and shipping, it is reasonable to say that the target Walmart.com grocery customer is essentially “people in the US who routinely buy groceries and prefer convenience and low prices”.
What are their customers’ goals?
A 2021 article on Walmart’s site for their third-party Marketplace sellers describes Walmart.com customers as budget-conscious shoppers who value convenient returns and shipping options, easy searching, and user reviews. They are loyal to Walmart and shop there at least once a week5.
How Walmart.com helps accomplish these goals:
Because Walmart’s shoppers are frequent/loyal shoppers who value low price and convenience, Walmart successfully targeted the grocery market with low-price, high-convenience strategies.
In 2019, they were awarded the Grocer of the Year by Progressive Grocer. Tom Ward, SVP of Digital Operations, said (emphasis mine) “Grocery’s obviously super-important to customers, and therefore it’s super-important to Walmart… I think people increasingly look to us to save them money, and more often now save them time as well. One of the biggest time crunches in any busy family’s week is grocery shopping. The easier we can make that, both in-store and online, for customers, the better. That’s where most of the focus on our innovation is being driven from online grocery, for sure.” 6
Specifically, Walmart.com helps to accomplish these goals by providing:
A wide variety of items at low prices with clear descriptions and customer reviews
Convenient and low-cost delivery, shipping, and returns.
Effective search capabilities
Recommendations at each step of the shopping process
Robust recommender capabilities are a key part of an effective UX, and we will explore the recommenders on Walmart.com in the next section.
Walmart.com Recommenders and UX: Buying Blueberries
On Walmart.com, the user encounters different types of recommendations at key points in the shopping process: first, on the product page, similar items are recommended.
In the example below of buying blueberries, the first recommendation is for other fruits “based on what other customers bought” (see image below). However, the fact that these are all just fruits indicates that this is likely a content-based system focused on properties of items only, not “customers” and any similarities they have to me. Also note that these items are filtered for those available to me: they are either in my store or they can be shipped to me, but there is no indication that they are looking at broader attributes of customer/item relationships.
Content-based systems to organize and understand relationships between items benefit both the user and Walmart: with many, many thousands of items, Walmart.com needs a robust way of categorizing and measuring similarity not only to drive customers to similar (perhaps more profitable) items, but to appropriately stock and promote items across geographies and seasons.
Later, in the cart, THREE separate approaches are used:
“Don’t Forget Your Usuals”: A list of products I’ve previously purchased multiple times but which are not in my cart. This is a simple intersection of the user and recent purchases of any type and does not recommend new items using any model.
Walmart.com app. Accessed 4/3/2024. “Recommended With Your Order”: While this appears similar to the content-based system that offered different fruits on the blueberry page, this is likely a mix of content-based recommendations based on items in my cart as well as a collaborative-filtering system based on users like me. For example, I had left some Plantone fertilizer in my cart and they are recommending other fertilizers: this is likely content-based.
But they also recommended granola, which is only loosely connected to fruits in my cart, so I believe they are referencing patterns of other users who also buy fruit.
These are all on-target recommendations and done seamlessly enough that it’s hard to say how they did it, which is ideal.
Walmart.com app. Accessed 4/3/2024. “Solar Eclipse Prep”: This is an interesting intersection of geography- and content-based information, and the only one that really misses the mark.
I am in PA, near the path of totality for the eclipse next Monday, and so Walmart.com serves up what it believes are relevant items:
- One set of eclipse glasses: The only item actually related to eclipse prep
- Multiple space-themed items like Moon Pies: These are a stretch, but related in a “fun” sense for kids, etc., so it works
- Outdoor items like lanterns and tents: This is a head-scratcher as it is a daytime event and does not require camping in the wilderness
- A box of raisins: This is a total miss; I believe the system is picking up the “sun” emphasis in the product name and description
Finally, after hitting the checkout button, I’m stopped by a screen asking me “Did You Forget Anything?” which again is a simple list of items I’ve purchased repeatedly but are not in my cart today.
Walmart.com Recommenders: Suggestions for Future Development
In conclusion, given Walmart’s massive investments in their technology, people, processes, and infrastructure, it is somewhat surprising to parse out what seem to be very simple recommender systems.
Since customers are looking for convenience, Walmart’s focus on putting my own purchases in my path does spur me to repeatedly buy a set of core items each week. Nothing could be faster for me to knock out this weekly chore.
However, I would recommend a few improvements:
- More collaborative-filtering: The app makes it very easy for me to order the same core groceries weekly, which is a major driver for me to buy my basics here each week. However, I am unaware of most items Walmart offers and every week, I buy non-core items at a local store. There are too many menus on Walmart.com for me to want to explore since I am here for speed: Walmart should refer to other customers like me to show me new items.
- More near-time, location-based recommendations: Walmart leverages geoanalysis and its huge data foundation to optimize in-store offerings for local preferences. With the very weak exception of the “eclipse supplies,” I do not see evidence of localized preferences online. I would like to see snow shovels when snow is in my forecast, not just pre-planned generic seasonal marketing.
1 McBride, S. (2021, January 11). Walmart Has Made A Genius Move To Beat Amazon. Forbes. https://www.forbes.com/sites/stephenmcbride1/2020/01/08/walmart-has-made-a-genius-move-to-beat-amazon/?sh=2e4b74da1119
2 Carmely, M. (2024, March 6). Walmart Target Market Analysis: Shopper Demographics and Insights – 2021-2022 Overview. Start.io - a Mobile Marketing and Audience Platform. https://www.start.io/blog/walmart-target-market-analysis-shopper-demographics-and-insights
3 Walmart U.S. sales share by product category 2023 | Statista. (2024, March 18). Statista. https://www.statista.com/statistics/252678/walmarts-net-sales-in-the-us-by-merchandise-unit/
4 Lin, Y. (n.d.). Online Grocery Market Share by Company (2024) [Jan ’24 Upd]. https://www.oberlo.com/statistics/online-grocery-market-share-by-company#:~:text=According%20to%20a%20recent%20analysis,steady%20from%20the%20previous%20year.
5 D. (2022, March 3). 9 things to know about Walmart.com customers. Walmart Marketplace. https://marketplace.walmart.com/9-things-about-walmart-customers/
6 Dudlicek, J. (2019, October 11). 2019 Retailer of the Year: Walmart Is Reshaping Grocery From the Ground Up. Progressive Grocer. https://progressivegrocer.com/2019-retailer-year-walmart-reshaping-grocery-ground