Welcome to the first installment of Greycroft Science, a blog post prepared in conjunction with Earnest Research. I would like to thank Lucy Wang, Masters candidate from NYU Courant, for her assistance in analysis and data science.

There is no more familiar topic than grocery shopping. Literally everyone has done it. You probably haven’t spent a lot of time thinking about the financial model of your local grocery store though.

A typical grocery store is a huge space, filled with perishable products, employing people who stand behind counters, stock shelves, clean, and manage cash registers. It is a low margin business, yet there are thousands of grocery chains in the US and some have been around for a hundred years. Most important from a VC perspective, the $800BN US grocery market has also been immune to the Internet, at least until now.

In the past two years there has been a flurry of investment activity, with companies like Plated, Blue Apron, Hello Fresh, and Instacart entering the online grocery business. These companies are backed by Kleiner Perkins, Bessemer, Sequoia, and (yours truly) Greycroft. We all know about the WebVan debacle two decades ago. So what is different this time around?

Prior to our Plated investment last year we spent months analyzing the grocery sector, trying to figure out what was going on. We decided to update and share a part of this analysis for an ongoing series. If you like what we are doing please leave a comment below.

How we did our analysis

We worked with Earnest Research, a Greycroft portfolio company, to develop a panel of 740,000 consumers who opted-in to sharing anonymous credit and debit card spending data. The resulting dataset is statistically representative of the entire US population, with confidence levels and intervals described below:

Panel Overview as of February 28, 2015 (click to expand)

This panel is not perfect, but it is pretty close.  It also has two advantages for analyzing start-up companies. First, the start-up companies above do not accept cash, so the panel is more representative for start-ups than traditional brick-and-mortar companies. Second, Earnest began collecting data before any of these start-ups were founded. That means we can see customers from first purchase onwards, which enables us to do things like track retention and observe spend differences between repeat customers and new customers.

As an aside, we also looked into restaurant delivery from Seamless, Grubhub, and Delivery.com, but we found little overlap between grocery spend and delivery of prepared food, so we saved that for a future project.

Sample insights

This panel allows us to easily calculate customer metrics. We included two tables below along with our analysis of the data.

Table 1.) Average Order Value (AOV) as of February 28, 2015 (click to expand)

Average order value is just one of many retail metrics, but it is a powerful factor in the grocery business because food costs are similar for each provider. For instance, the chart tells you that Whole Foods customers spend 20% more on each visit than Fresh Market customers. If you look at the financial statements you will see that this almost directly translates into gross margins: Whole Foods (NASDAQ: WFM) has 34% gross margins and Fresh Market (NASDAQ: TFM) has 29%. Instacart is the one outlier in our sample because they technically sell delivery and not food, and by our math have contribution margins in the 0-10% range (after delivery) versus 20-30% for a comparable grocery store.

Other thoughts on Table 1:

1.) Online grocery has significantly higher AOV than offline grocery. We looked at similar “online-to-offline” ratios across a dozen other industries and grocery had the most favorable ratio we could find. Returns are also nearly zero in online grocery.  The combination of these two factors is very promising for eCommerce providers.

2.) Instacart’s AOV stands out because it is so much lower than its competitors, Fresh Direct and Peapod. We dug a little deeper and looked at how much Instacart customers spend when they are shopping elsewhere, and, as it turns out, Instacart users have a much higher AOV than an average grocery consumer, just not when they are shopping on Instacart.

Our best guess is that one of two factors is at play – either Instacart’s mobile-first approach is less conducive to large baskets or Instacart’s business model discourages large orders by charging a mark-up on each individual item.

3.) The subscription box providers, Blue Apron and HelloFresh, have an almost identical offering but their pricing is clearly different.  They both start at 6 plates/box, for a list price of $60. HelloFresh offers large discounts for new users, resulting in new customer AOV that is almost 30% below Blue Apron. In spite of the discounts, HelloFresh lost market share to Blue Apron for most of 2014, and only recently surged back in January and February.

4.) Plated has a hybrid business model, offering 4 or 6 plate subscriptions, as well as a la carte items. Unlike the subscription providers above, every consumer gets a different box with Plated. You can see in the panel that this hybrid model has resulted in faster AOV growth, as customers tend to opt-into more add-ons over time. This was an important part of our investment thesis last July. Investors in Blue Apron will tell you that they get better margins by eliminating consumer choice, but that is more than outweighed by Plated’s extra $8/box.

Table 2.) Transaction Analysis as of February 28, 2015 (click to expand)

Thoughts on transaction data:

1.) The grocery market is seasonal. On average, customers spend 7% percent less in July and August, and 7% percent more in December and January. This is difficult to see in the above figures because February is a short month – even though there are 10% fewer days in February, spend figures were flat compared to August.

2.) US consumers appear to be indifferent to ordering groceries further in advance – Peapod and Fresh Direct require purchases to be made a day or more in advance, but they have almost the exact same purchase frequency as retail or Instacart. This raises a lot of questions about the future of on-demand grocery services versus a more traditional eCommerce approach.

3.) Just looking at tables #1 and #2, Peapod and Fresh Direct outperform every brick and mortar retailer in our sample set. Their customers spend more each month, order in advance, and require no consumer-facing storefront.  This likely results in higher margins, because advance ordering allows a company to anticipate demand, resulting in less food waste.

4.) The subscription providers all hover around two shipments per month across the user base. Plated has the fastest growing usage in the panel, which is due to the fact that they rolled out subscription products in the middle of the measurement period.

It is possible to use the Earnest panel to forecast monthly revenues, retention, gross margin, competitive wins/losses, and CLTV for every company in the sample. You can also track users before and after their first purchase, to see if new entrants are cannibalizing other companies in the panel. I did that for my own analysis prior to funding Plated. If you are interested in that level of detail you can contact Earnest for more information.

Conclusion

If you could build a “grocery of the future”, it would likely have the following characteristics: low food waste, limited working capital requirements, online ordering from every device you own, high margins, centralized operations, and delivery to your doorstep. In short, it would be the exact opposite of a conventional grocery store.

One key to achieving this vision is predicting what your customers will purchase in advance so that you can stock just the right items. This is impossible, of course, unless your customers give you a day or two forward notice. It is no surprise that Blue Apron, Hello Fresh, Plated, Fresh Direct, and Peapod all require users to order in advance. The best news in our analysis is that grocery customers are accustomed to planning ahead. I am curious to see whether AmazonFresh, which is the only “same day” online grocery in the country, will work.

A second option is to not be a grocery at all. You can just be a shopping cart on top of existing stores and rely on their infrastructure, like Instacart or Postmates. The only problem with that approach is margins, since you would be buying at retail and selling as “super retail” pricing. In the end, Instacart’s most valuable asset might be their data about what consumers are willing to pay for groceries and delivery, because that information can be used to price discriminate down to the end-user. Time will tell whether their idea is big enough to build a real billion dollar business or not.

As a New Yorker, I have had Fresh Direct for over a decade, so online grocery shopping is not revolutionary. I still shop at Whole Foods, order-in from Seamless, and cook Plated twice a week. In spite of all the market innovation, I have yet to see a grocery store go out of business in my neighborhood unless it was getting replaced by a more upscale grocery store. I do, however, see restaurants come and go all the time. My general feeling is that cooking is making a come back. And all the data I have seen suggests the same.