[Research] Instacart Pt.1

Faux-network or real network?

Not held.

---

1. Introduction & History

2. The Peculiarities of 3rd-Party Grocery Delivery

---

On this particular occasion I thought I’d take the opportunity to write about a company after it cratered on earnings.

---

It strikes me that the mental model here is that the success case for both DoorDash and 3P Grocery and Instacart is, basically, you're running the delivery sort of at break even on a product level and then you have a very profitable ads business on top of it.

Introduction & History

Instacart (CART) was founded in 2012 by Apoorva Mehta. Of Indian extraction and Canadian nationality, Mehta took the inspiration for CART from his arduous bus commute to get groceries in the Canadian Winter. As a back-end logistics engineer for Amazon in Seattle, Mehta dreamed of starting his own company. In 2010 he quit and moved to San Francisco. After a string of failed startups, he finally ideated the concept for CART and promised to not go to the grocery store until the code for it was completed. In late 2012, he secured admission to Y-combinator by having the CART app deliver a six-pack of beer to YC partner Gary Tan after missing the admission date by two months. The rest, as they say, is history.

The concept that picked up product market fit was elegant:

Instacart is a product where you can order your groceries and get them delivered to your door within 1 hour. What’s interesting about this is how we actually make this happen. Instacart is an entirely software company. This means we don’t have any warehouses, no trucks, and we don’t hold any inventory. So when you order your groceries, we have one of the thousands of personal shoppers in our network pick up your groceries from stores such as Wholefoods, CostCo, and Safeway, as well as many others, and have them brought to your door within 1 hour.

Apoorva Mehta

Originally this was a purely mobile experience. Mobile, was, and likely still is, what enables this business to exist, period. CART itself was preceded by a number of Web1.0 concepts in grocery delivery, most infamously Webvan. Webvan stood in stark contrast to CART: WV built their own fulfilment infrastructure, and owned their own trucks. They went the capital intensive route and inevitably went bankrupt in 2001, owing no less than $800M. The advent of the iPhone allowed for a business that was both networked (the efficient linking of personal shoppers who were willing to pick, pack, and deliver groceries, and people who didn’t want to go shopping themselves) and could call on the capital of others (deliveries were made in the shopper’s car, and grocery purchases and delivery fees were paid by customer ahead of time).

The first personal shoppers would text Mehta after their orders were completed. In fact, the personal shopper was an unusual hack - from the retailer’s perspective it was just another person walking into their physical location to buy groceries. Long time CART execs fondly refers to this as ninja shopping. From CART’s perspective, this was the beginning of aggregating the demand side of the equation without necessarily needing to consult the supply side. The only hitch with side-lining the retailer (initially at least) was the lack of adequate SKU information to display to customers. This challenge was obviated by CART actually physically going into all relevant retail outlets and purchasing each individual product, photographing it, and uploading it to the app. This was the beginning of an unusual competitive advantage - they became the only centralised data depository of SKU information across retailers. What’s more, in time they would get even better information on shelf inventory than the retailer themselves(!). Even today CART enjoys the largest data edge in this regard. Proper imagery and accompanying information doubled sales for the first retail outlets they onboarded manually (later with their knowledge and consent). Once the demand side was aggregated (i.e. customers were clamouring for the products) it would be much easier for CART to approach retailers on a partnership basis.

CART first rolled out in San Francisco and moved into adjacent markets over time. In 2013, they began delivering alcohol in locations where this was legal. In the same year they introduced Instacart Express (later renamed Instacart+) as an annual membership option. In 2017, they expanded into Canada. Retailer partnerships became commonplace early on in their existence as a company.

The original iteration of monetisation is interesting. This took the form of an optional gratuity of between 5-10% of the older size. Over the first five or so years, the company toyed with different setups. At first it was purely optional, and the gratuity option had to be selected. Later it was made a default setting. Today, fees are mandatory and are subject to an array of variables (convenience, retailer, promotional periods etc). The point here is that a business like CART could only accrue a market position like it did, from scratch, in the free money world of the 2010’s. Replicating its initial competitive advantages is hard enough, funding those in a world where the cost of capital is significant higher than 0% is quite another.

Typical order sizes see fees of between 10-13% of the order size.

Like so many things, the Pandemic was a watershed for the company. CART’s sales increased by a factor of 5 in 2020! This was followed up with a very significant uptake in personal shoppers, although nowhere to the same extent. The post-Pandemic era saw CART list via a normal IPO in 2023, a very rare occurrence in recent years. Apoorva Mehta stepped down as CEO, and subsequently from the board, and was succeeded by former Meta Platforms executive Fidji Simo.

If we speed up to today, CART is a lot more than a grocery delivery app. Obviously, underlying their operations is grocery delivery. Today CART has over 600,000 personal shoppers in North America. They partner with 1,500 retailers, provide delivery from 85,000 grocery outlets, and work with 6000 consumer packaged goods (CPG) companies. 13.7 million users are active on the app every day. Layered on top of the key consumer experiences (on both desktop and mobile) is an advertising business. Thematically, the advertising business is not dissimilar from how retailers typically make their money - the so called ‘shelf fees’. CART’s digital real estate is valuable and CPG companies pay for prioritised listings. The digital advertising experience is, of course, a significant step up from the form of traditional retailer advertising alluded to before. The starkest difference is that the advertising outcomes are measurable. Underlying CART’s relationship with the retailers is a number of B2B products. This includes inventory management tools, advertising products for the retailers website, as well as a white-labelled e-commerce front to power these direct bookings.

Conceptually, this is something of a four-sided marketplace - although the lateral integration across constituents is unique and underappreciated. The value here concentrically builds out from the delivery service, without it their advertising business and inventory management tools likely wouldn’t be worth much. In the value chain there is, of course, the customer who orders products, the retailer who stocks these products, the personal shopper who delivers the products, and then there is the CPG company whose products are being sold. In this the retailer is a net beneficiary of the marketplace, the delivery connection is very modestly monetised, and the advertising dollars generated through the CPG’s is the most profitable. Delivery is a notoriously difficult business and it requires some analysis.

The Peculiarities of 3rd-Party Grocery Delivery

As mentioned above, a strange collection of characteristics exist exclusively in North America in grocery. Firstly, CART’s offering is capital efficient - more capital intensive models have failed, and even the largest retailers themselves have achieved various levels of mediocrity in trying to do delivery in-house. In the main, large ticket grocery delivery needs a ridiculous amount of network density to become worthwhile doing. You can think of most grocery sales being between a 1-2% margin business for even the most scaled retailers. When you add on the delivery costs to a fairly dispersed geographical area, customers end up paying through the nose. Home delivery was a mainstay of the smaller, family-run grocer of the past. It was largely done away with by the stalwarts who came to dominate the American economy post-World War 2.

Even these retailers have typically ended up in some sort of partnership with CART out of pure convenience and the additional shopping traffic driven to their outlets (Whole Foods being an interesting exception we will discuss later). The personal shopper side of the network also obviates the fixed cost nature of in-housed delivery. Those personal shoppers who work on a gig-like basis rely on tips to make decent money ($30 US/hr seems to be the goal for most). They drive their own cars, purchase their own gas, and they (largely) work their own hours (or work for the algorithm, as they like to call it).

Instacart has, like other gig-economy platforms, come under the ire of employment regulation. They have limited offerings for personal shoppers to enjoy part-time employment where the can enjoy the typical perks of customary W2 labour (401K contributions, healthcare, etc). This arrangement doesn’t necessarily change the spectre of the work - but in some cases it may actually cap the actual hours workable in a week!

Secondly, North America, and the US in particular, enjoys a stunningly diverse range of retail outlets. My home country Australia, for example, has a grocery oligopoly because of the subscale nature of the market. The applicable grocery delivery network would be subject to relatively few retailers (with only two having significant scale) and a much less dense collection of metros, whose economics subsidise delivery in less populated geographies. Home delivery start-ups have failed spectacularly, and the service is really only reserved for wealthier households. North America represents one of the best markets for this kind of setup. The pure scale of the parties on each side of the network is essential. It took millions and millions of deliveries for CART to reach profitability (and even that is a very significant achievement) in delivery alone which belies the challenge experienced by any individual retailer in doing delivery themselves. At first glance, CART appears as an outsourced delivery service with business upside rather than the threat of disintermediation to the retailer. Like parasite offering more positives than drawbacks. History would indicate that one can either run a profitable retailer OR a profitable delivery network, but not both. This brings us to the acute issue of Amazon, and the Whole Foods saga.

In mid-2017, Whole Foods - then representing about 50% of CART’s business - was sold to Amazon for $13.7B. This was something of an existential crisis for the CART, but it was also a call to action. Senior management made the definitive decision to deeply embed themselves with their remaining retailers, as well as strike up partnerships with anyone else who wasn’t with them already. The existential nature of “Amazon Threat” has turned out to be less existential than first thought.

The post-facto analysis of Amazon’s move into grocery distribution seems to be highlighted by the fact that most grocery retailers are highly sceptical of Amazon as a potential distribution partner (all they need to look at is how Amazon treats their 3rd party merchants elsewhere to see what writing could be on the wall), and Amazon isn’t in a position to buy every bricks and mortar retailer in North America. In any event, Amazon’s threat to CART is vicarious: They are threatening to CART to the same extent that they are a threat to the entirety of bricks and mortar grocery retailing. In this respect, Amazon has had very mixed success. The nuances of grocery are remarkably different from other forms of commerce, internet empowered or not. While they undoubtedly have the scale to compete vigorously, they have not been a category killer - and in many cases they aren’t able to provide the same level of convenience, or selection. Both are critical. Prime’s promise of grocery delivery is within 2 hours - CART’s is half of that. This goes back to the original point - the asset-light and platform nature of CART’s offering somehow offers a more convenient (and I’d argue that is the determining vector here) option to a customer than a competitor who is vertically integrated. I’ll say candidly, this aspect is very unique, but admittedly not insurmountable on a long enough time frame.

Grocery is only about 14% penetrated online and it represents over $1.2T per year in sales. Where this could go over the long-term is anyone’s guess but most informed speculations seem to be over the opinion that penetration will at least double in the vertical.

Likewise, both Uber and Doordash have been actively investing in their grocery operations but while they enjoy some of the same scale and network advantages that CART might, they have found significant trouble in migrating the customer experience, and especially the delivery/personal shopper experience from small to large basket. These limitations might not seem especially obvious on the surface level, but even seemingly superficial differences can make transplanting services extremely difficult. Doordash’s business, for example, is in what most would understand as “small basket” - imagine getting dinner delivered for yourself and your partner. At most there might be 6 or so individual items in a delivery (no offence if your order is typically much larger!). The restaurant creates the order, packages it, and hands it over to a delivery person. The discretionary nature of the delivery relies almost solely on a convenience factor, which explains why customers are so willing to pay inflated prices for their orders. In the vast majority of cases this delivery will be a 1-1 connection. The delivery person hands you your order at your home. Voila!

Grocery is a categorically different beast. Typical orders on CART are over $200 (a typical customer life cycle is they begin with much smaller baskets, which grow over time). Baskets can have over 100 items in them. Importantly, in the case where a personal shopper picks the items, discernment is required. They may need to check eggs for cracks. They may need to know where a particular product is located in a supermarket. Features like substitution (where a particular item may be sold out and a roughly equivalent item can be substituted for it) was only introduced by DoorDash a few years ago. Only after they have picked and packed the delivery can they actually deliver it. This process has been somewhat obviated lately as more grocery partners have opted for having their own staff pack orders for shoppers. This undoubtedly brings the shopper experience closer to that of a Dasher, but not in every particular case. They also face the uphill battle of trying to replicate the SKU database that CART aggregated some time ago. This is to say nothing of the deeper integrations that CART has with retailers already. The most common feedback for both customers and shoppers online is: go with CART if they are looking for deliveries over 10 items.

DoorDash for instance has decided to actually build out its own physical infrastructure: DashMart’s. This kind of haphazard move into physical infrastructure is a tell, like in poker. It’s a weak hand. They’re, at this time, probably not built for it.

There is something to how customer basket sizes develop over a customer’s lifetime. Just as going to the supermarket physically is habit forming, it’s also true that use of deliveries is habit forming. My bank account can attest to this. As mentioned above, initial basket sizes are smaller and as a customer becomes more accustomed to the service they order more regular sized deliveries, at more regular frequencies. This actually seems like bit of an in for the likes of DoorDash who are customarily smaller ticket. Migrating that user experience, with the same level of service is the open question that is left to be answered. The fact that they haven’t been able to do it easily says something.

From the retailers perspective they have no particular interest in exclusively utilising any one provider, they are, after all, providing a place for the providers to pick up inventory. Regularly bad service could have negative associations with their brand but that seems not determinative. Having their wares discoverable on a single forum likely isn’t taking away competitive advantages vis-a-vis secrecy - retail is the most transparent market in the world. The obvious threat here is from large retailers offering their own delivery. I’ve already covered why this is difficult (high fixed costs, lack of network density, difficulty in offering equivalent service) but at this stage it’s probably true that if a retailer were to exclusively utilise their own delivery they would actually be hurting their own sales, and they’d likely have years of margin declines for very little long term return.

At least by their own reckoning, CART is not just a delivery company (and certainly by its financial statements too). To be sure, the entire franchise or network (a looser application of those terms than usual) is built upon delivery, but if that’s all there was than this would be a weak business. There is evidence, admittedly, to back up the claim that they are actually grocery technologists (I cringe writing that). If anything would sufficiently differentiate themselves from the onslaught of competition that they will inevitably face forever, it would the array of tools they provide to retailers. I also thinks elements of the advertising business could have interesting lock-in on a long enough timeline… but that’s for the next entry.

---

Larry.