DoorDash, Instacart, and Shipt: A Tale of 3 "COVID Bumps"

There has been a lot of debate swirling about Instacart lately. News emerged that the company had been in discussions about potentially being acquired by DoorDash. That and management changes have left many wondering just how well it is doing, especially now that peak pandemic-induced demand is likely behind us.

There has been much less talk about Shipt, another rather well-known delivery company now owned by Target. But because it had come up in commentary after I'd discussed the issues at Instacart, I figured it would be interesting to dig more into how they have been doing as well.

In this note, I'll be focusing on the following questions:

1. How is Instacart doing from a revenue perspective (and just how big is it)?
2. How does Instacart's topline and repeat purchase performance compare to DoorDash's?
3. How is Shipt doing from a revenue and repeat purchase standpoint?

I went about answering these questions using data from Earnest Research, one of the leading credit card panel data companies, who had served as the data source underlying my paper with Elliot Oblander on the impact of COVID on the restaurant delivery category. I had data from them running from January 2016 through June 2021, providing us with a fairly current view of purchase trends. As those who follow me know, I believe this sort of purchase data is highly informative when used correctly, providing us with a lot of granular insight into the performance of companies like these. We had showed in that paper that this panel spending data explains literally 99% of all the variation in "gold standard" gross order value SEC disclosures, and below, I'll provide a bit more evidence of that... and use this data to triangulate my way back into what Instacart's GOV is.

For those who are also a bit short on time, these are the main conclusions:

1. Instacart has held onto less of its pandemic-induced gains than DoorDash, with y/y spending declining the most recent 3 months in a row...
2. ... but Instacart had also gotten a bigger "COVID bump" than DoorDash did, which may partially explain the relatively weaker performance in recent months.
3. Repeat purchase activity is one of the other culprits -- it is significantly weaker at Instacart than at DoorDash. The gap is upwards of 50% between the two of them.
4. Instacart did $20 - 30 billion in GOV in 2020 (my point estimate would be ~$23 billion).
5. The wheels appear to be falling off at Shipt. Sales now appears to be below the pre-pandemic baseline trend, with sales in the most recent couple of months being down upwards of 66% (!). Repeat purchase patterns, while strong in early months and for older cohorts, are worse than those of both Instacart and DoorDash as tenure goes up (and even more so for the younger cohorts).

For those interested, I'll unpack these main conclusions below.

Instacart: topline performance

Instacart's sales have been slowing lately. Here is a chart of Instacart's total GOV in the panel over time:

GOV had been growing very steadily, then spiked during COVID. Sales growth, which had been vacillating between 70% and 200%, jumped to a peak of ~600% in July 2020 before settling down. Instacart's GOV is still clearly well above their pre-pandemic trend, but growth has been slowing. In fact, growth has slowed so much that Instacart's year-on-year change in sales has been negative the past 3 months in a row:

Of course, part of this is to be expected -- incoming CEO Fidji Simo described this as their new "resting heart rate" in a recent interview -- but not all companies have been seeing this sort of weakness. Sales trends at DoorDash have remained remarkably robust, even in recent months. Below is a comparison of Instacart to DoorDash along the above measures:

DoorDash grew more strongly than Instacart pre-pandemic, less strongly during the fever peak of the pandemic, but has been holding onto that sales growth more strongly as the pandemic has waned (relatively speaking). Sure, growth has slowed at DoorDash, but it is still growing at around 33% year-on-year in the panel data.

Why would this be?

Repeat purchasing at DoorDash and Instacart

It is in no small part due to the durability of DoorDash's revenue being better than at Instacart. At the risk of information overload, this is a graph of the proportion of DoorDash customers that are active in future months, by cohort:

I know, it's a lot of lines. Allow me to explain. A customer is born when he/she makes his/her first purchase, so all customers must be active when they are first born (hence all lines starting at 100%). After that, not everyone will buy. What we see from this graph is that by and large, after a cohort is formed, the proportion of the cohort that is active falls to around 22-23%, but then picks up again, rising by about 1 percentage point per year thereafter.

These patterns were relatively stable across cohorts.

During the pandemic, these cohorts all got more active, with the proportion of each cohort remaining active jumping up to around 33%.

How about Instacart?

Here's the corresponding activity graph for Instacart:

While it may not seem obvious given how much is going on in this graph, these repeat purchase patterns are not nearly as good as at DoorDash. Basically, while repeat purchasing is relatively comparable between the two companies a couple of months after a cohort is born, the proportion of a cohort that is active generally tends to fall over time at Instacart, while it tends to rise over time at DoorDash. There is a healthy gap that just tends to get bigger over time.

Here are a couple of charts that compare these activity cohorts directly between the two firms (all other cohorts show the same substantive trends):

It makes a big difference for revenue durability when 30% of a cohort is active, versus 20% -- at the risk of stating the obvious, that is literally 50% more repeat purchasing.

I would have intuitively expected more repeat business at Instacart, given how "regular" the underlying buyer behavior in the category is. They don't appear to be that effective at converting adopters into online grocery shoppers.

Just how big is Instacart from a GOV perspective?

I really enjoyed Laura Forman's recent article on Instacart, but one thing stood out to me - there is very little in the public domain about just how big Instacart from a GOV perspective. From Forman:

"In part because of their larger average basket sizes, groceries as a core business relative to restaurant food also make for very different unit economics. Instacart last year said it did $1.5 billion of revenue but handled “tens of billions” of dollars in gross transaction value. While not a perfect comparison,?DoorDash?generated almost twice the revenue last year, handling roughly $25 billion in gross order value for its largest business."

This begs the question - just what exactly does "tens of billions" mean? Are we talking $20B big, or $80B big?

One way that we can triangulate our way into this is again through this credit card panel data. At the risk of re-stating the obvious, the Earnest data has been very consistent with DoorDash's SEC disclosures. Here are the annual spending figures in the Earnest panel, in DoorDash's SEC disclosures, how the ratio between the two (and the corresponding y/y percentage changes) have evolved:

Bottom line, the panel represents about 1.3% of the overall population (using Earnest's Consistent Shopper Panel).

So one thing we can do is take that same 1.3% figure and gross up Instacart's panel spending figures, to estimate Instacart's GOV here in the US. Given how large the footprint is of both firms, the fact that both are for all intents and purposes US-only, and the fact that our coverage of both firms' transactions should be very high (because they are both digital-only), this is a very reasonable place to start. Doing so yield's the following figures for Instacart:

If Earnest's panel represents 1.3% of the overall population, then Instacart did about $23 billion in GOV in 2020. While this may not be perfectly accurate, the percentage is almost surely in the 1.1-1.5% range, implying that Instacart's total GOV is in the range of $20-28 billion. This is on the lower side of what one might think when one hears "tens of billions," which when you think about where this statement was coming from, makes total sense.

Laura, this one's for you!

If Instacart's 2020 GOV was $23B, and its revenue was $1.5B, its revenue was about 6.5% of GOV. Over the same period, DoorDash generated $2.9B in revenue, which represents about 11.7% of its $25B in GOV.

What is going on with Shipt?

I like Shipt. I use them every once in a while, and while they are pricey, I have always had a good experience with them. Which is why I was so surprised to see just how poorly they seem to be doing no matter how I cut the data.

This is a chart of overall spending at Shipt:

This is significantly worse than at either Instacart or DoorDash. In fact, their GOV has fallen below their pre-pandemic trend. Year-on-year spending, which had been relatively strong through 2018, fell to just 10% in January 2020 before exploding upwards, but in May and June 2021, y/y sales have fallen by about 66%:

This is quite a steep drop.

Naturally, one might think that this could just be a case of Target turning off the paid customer acquisition budget. But even the repeat purchasing trends have been steadily deteriorating over time.

One way to see this is by looking at the average spend per customer by acquisition cohort. I show this below:

Each acquisition cohort is just a bit worse than the one that precedes it. Customers born in 2017 were doing around $200 in spending in their first month, with their spending falling steadily thereafter. Cohorts acquired in 2019 were doing more like $160 in revenue in their first month, about 20% less than their older counterparts.

Another way we can see this is by looking once again at the proportion of each acquisition cohort that is active, as we had done with Instacart and DoorDash. This is the resulting chart:

Cohort activity is actually relatively good for those older cohorts (absolutely and relative to Instacart and DoorDash), but just tends to get worse and worse, both over time and across cohorts.

To make the cross-company comparison easier to see, we can revisit that March 2017 cohort that we had looked at earlier, and add Shipt to it:

Shipt's cohort activity is significantly better than both Instacart and DoorDash for lower tenures, but as the cohorts age, Shipt moves into last place. And notice - almost no COVID bump at all. That blue line moved up a little during COVID, but resumed steadily falling shortly thereafter.

As we move to more recently acquired cohorts, the picture looks even bleaker at Shipt.