A few days ago, researchers at Cornell published this fascinating study around differences in online and offline grocery shopping baskets — and I haven't been able to stop thinking about it ever since.
The study documents four important findings after analyzing data on nearly two million offline and online (Instacart) shopping trips across 4388 households in the USA:
1. Basket variety in online grocery shopping is lower
On average, online grocery shopping baskets exhibit significantly lower variety in terms of unique food categories and items purchased than brick-and-mortar baskets.
In other words, consumers shop a higher variety of products when browsing a retail store versus while ordering from Instacart.
Specifically, the researchers found that consumers tended to purchase 9.6% fewer product categories on Instacart and 14.1% fewer products on average, while shopping on Instacart as compared to shopping at offline retail stores.
Possibly, when you go to the grocery store you buy whatever you feel like. You discover stuff, so there's more variety in your shopping. But when you buy on Instacart, you just buy the same things as last time, so it has lower variety.
And yeah, if you're thinking that perhaps it is so because fewer categories of products are being sold online versus offline, the researchers made sure that they were only accounting for products that were available on both online and offline channels and being sold by the same retailers.
2. Online grocery shopping exhibits less diversity between purchases
The researchers found that Instacart trips are 27% more similar to each other than offline trips within the same household when comparing categories. In other words, Instacart shopping trips have twice as much item overlap between any two successive trips within the same retailer as compared to offline shopping trips.
Now, you might say that perhaps users are not comfortable with using online shopping apps, so they end up discovering fewer products as compared to offline stores.
What the study found is: Although households with more experience using the online channel tend to show less pronounced differences in variety and similarity, the gap persists even among the most experienced users!
This indicates that for planned online shopping platforms, the barrier to entry for new products may be higher, as consumers tend to repeat purchases. The usual winners keep winning.
3. Instacart shopping baskets had fewer fresh vegetables and impulse purchases
The researchers found that Instacart shopping baskets had 13.6% fewer fresh vegetables compared to the brick-and-mortar shopping baskets. At the same time, they also saw fewer impulse purchase categories such as candy (7.1% fewer), bakery desserts (5.9%), and savory snacks such as chips (4.7%).
The researchers conjecture that the lack of fresh vegetables in online shopping baskets may be due to the fact that buyers don't trust their Instacart shoppers to pick the best vegetables for them; they consider picking fresh vegetables as something they would best do themselves, in an offline setting. But the study does not supply any answers in this regard.
(There's also an important caveat here that I will discuss further down the piece, at length.)
4. Online shopping meant fewer purchases overall
The researchers tried to see if consumers were making up for their lower basket diversity on Instacart with additional trips to buy those items they would've bought if they were in a retail store. Turns out, households did not compensate for the fewer fresh vegetable and impulse purchases via alternative or additional trips.
You cannot establish trust in the findings of any research without understanding how sound the methodology behind the research was.
For this study, the researchers identified what they call “characteristic trips” for both Instacart and offline shopping visits. They then compared these two sets of characteristic trips to come up with their findings.
“We observe close to two million shopping trips for our sample of 4,388 households. These shopping trips are comprised of grocery trips in which a given household makes purchases that reflect regular restocking of commonly used household grocery items; and non-regular (intermittent) shopping trips, such as fill-in trips or runs to convenience stores or gas stations. Our objective is to examine the shopping histories of each household to identify their routine restocking grocery trips, irrespective of the channel. In other words, we aim to identify those Instacart trips that serve as likely substitutes for a BM trip. We refer to such trips as characteristic trips.”
Here's the rationale behind identifying a trip as a characteristic trip or non-characteristic trip:
“Panel (A) shows density plots of the spacing (in days) between two successive trips made by a household. Characteristic trips have significantly longer spacing between successive trips compared to non-characteristic trips (206% higher; t = 82.3, p < 0.000). This suggests that non-characteristic trips may represent unplanned runs to the shops with a higher sense of urgency.
This assertion is further supported by the data patterns depicted in Panel (B), where we see that the average number of items in a basket is significantly lower for non-characteristic trips and that the basket size of noncharacteristic trips does not vary with the total frequency of household trips.
In contrast, the size of the basket for characteristic trips varies significantly with the frequency of trips: households that shop less frequently have larger characteristic baskets, and the number of items in the characteristic baskets decreases as households increase the shopping frequency. This might suggest that non-characteristic trips satisfy some urgent needs of the household, while characteristic trips tend to satisfy the less urgent grocery needs that can be achieved with varying frequency.”
To put it simply, the researchers categorized characteristic and non-characteristic trips based on the frequency of each trip — more periodic and less frequent for characteristic trips, indicating prior planning and scheduling, and the basket size — higher for characteristic trips, again indicating some level of planning on the part of the household.
Non-characteristic trips, on the other hand, were trips with an erratic frequency and lower basket sizes in general — indicating more urgent and unplanned needs.
After creating these two sets, the researchers applied some fancy math and machine learning — the details of which I leave up to you to decipher. Essentially, they calculated distances between Instacart and brick-and-mortar trips in a 100-parameter Euclidean space and found out that Instacart trips were more bunched up in space as compared to brick-and-mortar trips — indicating that the former are way more similar to each other than the latter.
Shopping Fast and Slow
The very segregation of trips into characteristic and non-characteristic trips raises an important doubt. And my doubt stems from this:
I've heard from many D2C founders and marketers that quickcommerce platforms like Swiggy Instamart, Zepto, and BlinkIt are some of the most effective channels for launching new D2C brands.
Because the propensity of the consumer to try new products on these channels is very high.
Now, at first glance, this might sound quite antithetical to the findings of the study we just discussed: Instacart basket diversity is generally lower than retail store diversity.
How do we explain this conflict, then?
The answer lies in how the study defines characteristic trips, which by definition, ignores unplanned and impulsive orders that quickcommerce is known to inspire.
Instacart promises same-day delivery with an average order arrival time of 2–5 hours; quite longer than quickcommerce turnaround times of 10–30 minutes. This is something this study does not seem to account for, as far as I've understood.
And this crucial bit of nuance matters. Here's how.
Now, to understand why unplanned orders lead to more impulsive purchases and a higher propensity to explore and try new products, here's a Harvard study that asked this rather relevant question:
How do people’s preferences differ when they make choices for the near term versus the more distant future?
From a field study of an online grocer, the researchers found that people act as if they will be increasingly virtuous the further into the future they project.
Researchers examined how the length of delay between when an online grocery order is completed and when it is delivered affects what consumers order. And they found that consumers purchase more “should” (healthy) groceries such as vegetables and less “want” (unhealthy) groceries such as ice cream the greater the delay between order completion and order delivery.
In other words, consumers spend less and order a higher percentage of “should” items and a lower percentage of “want” items the further in advance of delivery they place a grocery order.
Fascinating, isn't it?
In Daniel Kahneman's worldview, the more planned an order, the more of your rational System 2 thinking takes over, while the more unplanned and spontaneous an order, the more you tend to engage in System 1 thinking and buy on impulse. This is buying behaviour that makes quickcommerce channels so effective for launching new D2C brands.
(System 1 thinking is a near-instantaneous process; it happens automatically, intuitively, and with little effort. It's driven by instinct and our experiences. System 2 thinking is slower and requires more effort. It is conscious and logical.)
If you find this hard to digest, here's a pertinent analogy in the offline retail world that puts things into perspective.
Have you wondered why some items like mints, chewing gum, chocolates, hand sanitizers, sodas, and stationery items are kept at the checkout counter?
The simple reason is that the checkout line is one of the most highly visible places in a supermarket, which guarantees a look even if you have no intention of buying these products at all.
These products are called impulse goods. They are cheap, so can be bought without any kind of planning and without much risk of loss involved. They also are regular consumption items, so they do not need a special reason to be bought. They are purchased because you may just want to, as an impulse, without diving deep.
This is also the nature of a spontaneous order placed on Instamart or Zepto during a late-night Netflix binge. And it is the very nature of 10-minute deliveries that makes this impulsiveness possible.
Because think about it:
If your order took a minimum of 24 hours to arrive at your doorstep, you will tend to plan your purchases, enquire from others in your family if they want something too, make a checklist in your notes app, and then place an order based on that checklist. As the delivery times are high, missing out on ordering an essential proves to be an expensive mistake.
So, while shopping on BigBasket and Amazon, you naturally tend to engage in more System 2, rational thinking. You tend to order more of what you need and are less inclined to divert away from your premade checklist. And this premade checklist eliminates the possibility of impulse purchases.
Quickcommerce, on the other hand, is designed by nature for spontaneity and impulsivity. This is something that the Cornell study doesn't cover.
Serendipitous discovery in the real world
Now, let's come back to the Cornell study and try to understand why variety and diversity in planned online shopping might, on average, be lower than that in brick-and-mortar retail shopping.
For this, I will need you to imagine yourself waltzing around the aisles of a supermarket or local departmental store. You will immediately realize that the sheer number of items in your peripheral vision or field of view is way higher than what you see on your dainty smartphone screen. This makes the cognitive cost of discovery much lower in a departmental store than it is in a smartphone app.
Also, you don't know what you don't know, right? You can't voluntarily search for a product you do not know exists in an app's search bar. But in an offline setting, you just come across new stuff in the process of buying what you came in to buy.
Apps today try to aid product discovery, sure, but then there's the usual product design conflict between helping the user discover more products and prolong order placement times, or helping them buy what they're already used to buying and shorten order placement times. What ends up happening is essential product categories that we are used to buying periodically do not experience or undergo any kind of change. We tend to keep buying more of the same thing when it comes to household essentials. Convenience wins over exploration.
Even the propensity to explore or try alternatives is lower in an online setting. In an offline setting, you may still have a shopkeeper recommend you a better alternative that has hit the shelves recently, but an app suggesting new alternatives can never compete with a real person in terms of the palpability and effectiveness of the recommendation.
In this beautiful essay, author Leon Wieseltier mourns this loss of serendipitous discovery in online commerce and chalks it up to a difference between “searching” and “browsing”:
“It is a matter of some importance that the nature of browsing be properly understood. Browsing is a method of humanistic education. It gathers not information but impressions, and refines them by brief (but longer than 29 seconds!) immersions in sound or language. Browsing is to Amazon what flaneurie is to Google Earth. It is an immediate encounter with the actual object of curiosity. The browser (no, not that one) is the flaneur in a room. Browsing is not idleness; or rather, it is active idleness — an exploring capacity, a kind of questing non-instrumental behavior.
Browsing is the opposite of “search.” Search is precise, browsing is imprecise. When you search, you find what you were looking for; when you browse, you find what you were not looking for. Search corrects your knowledge, browsing corrects your ignorance. Search narrows, browsing enlarges. It does so by means of accidents, of unexpected adjacencies and improbable associations. On Amazon, by contrast, there are no accidents. Its adjacencies are expected and its associations are probable, because it is programmed for precedents. It takes you to where you have already been—to what you have already bought or thought of buying, and to similar things. It sells similarities. After all, serendipity is a poor business model. But serendipity is how the spirit is renewed; and a record store, like a bookstore, is nothing less than an institution of spiritual renewal.”
And I wholeheartedly agree with this analysis. In fact, I have written about how Kirana Stores aid local product discovery and consumer preference discovery.
I have discovered more books in Blossoms on Church Street than I have on Amazon. The very nature of an online medium doesn't inspire discovery like spending an afternoon in a bookstore does. On Amazon, you go to buy what you know you want; in a bookstore, you go to buy what you didn't know you wanted or needed. You go to explore and discover.
Online shopping does kill serendipity.
But does quickcommerce bring it back — at least for consumables, if not books?
This is a question I'd like you to think about, considering how our buying behaviour changes when we longer delivery times force us to aggregate and plan versus how 10-minute deliveries allow us to dive into the shopping experience without planning anything.
This also ties into the difference in average order values, frequency of orders, and total order volumes on both slow commerce and quickcommerce channels — but I leave it up to you to think about these things.