Red Apron

A Few Concerning Questions from Blue Apron’s IPO Filing

Blue Apron recently filed their S-1 IPO documents, which you can access here. It’s a fascinating look into the economics of the monthly subscription box space, particularly food, which has its own unique challenges.

After quickly looking through the document, I’m left troubled by what seems to be intentional shrouding of the true unit economics of the business. There’s a lot of detail in there, but sometimes what’s not said reveals more about the health of a company…

What’s the true Customer Acquisition Cost (CAC)?

This is something that has been highlighted by other, smarter business analysts, but I’ll throw my hat in here too because it’s just so glaringly important.

Let’s take a look at the only chart Blue Apron provided regarding CAC:

At first glance, the above chart implies that acquiring customers only costs $94. However, this is over a 3-year timespan — if you go back to their P&L, over the past 12 months (March ’16 — March ’17) Blue Apron spent ~$179 million to add ~387,000 active customers. This would imply a whopping $463 CAC to acquire a user that’s still active in March ’17 from the past year.

A VC at NextView Ventures pointed out that “active customers,” as defined in the S-1 (to arrive at the 387,000 customer metric), is not a true indication of new customer adds:

Some have suggested efficiency overall appears to be getting worse. For example this Recode piece incorrectly infers Blue Apron has recently been spending >$400 to acquire each new customer. The error is that “Customers” as reported in the S-1 is basically active customers (e.g. how many people ordered in a particular quarter / year), not cumulative customers added.

Which is true, yes, and it’s unfortunate that Blue Apron doesn’t disclose CAC by cohort.

What we can do is figure out the net adds in the whole ‘14-’17 period to get a holistic sense of the metric: adding up total marketing from ‘14-’17 (~$270m) and dividing it up by the average CAC ($94), we get to around 2.9m new customers in the entire period. So this probably means trailing 12-month CAC is probably lower than the $438 calc, but the flip side of that is retention is probably bad. CAC is also likely going up very quickly (typically early adopters are cheaper to acquire for any subscription business), and it’s frustrating that Blue Apron didn’t disclose more detail.

What’s “gross revenue,” and why is it not reported?

Blue Apron reports a top-line “net revenue” metric:

We deduct promotional discounts and customer credits and refunds expected to be issued to determine net revenue. [emphasis mine]

What? Are there additional “promotional discounts” which are not reflected in the P&L? Why would that not be reflected in marketing expense? So what exactly IS marketing expense? From the S-1:

Our marketing expenses consist primarily of costs incurred to acquire new customers, retain existing customers and build our brand awareness through various offline and online paid advertising channels, including television, digital and social media, direct mail, radio and podcasts, and email.
Also included in marketing expenses are the costs of orders through our customer referral program, in which certain existing customers may invite others to receive a complimentary meal delivery, as well as costs paid to third parties to market our products. The cost of the customer referral program is based on our costs incurred for fulfilling a complimentary meal delivery, including product and fulfillment costs. [emphasis mine]

Are there additional 1st party product promotions beyond the customer referral program? The first thing that comes to mind are those extra credits / coupons most on-demand companies (Uber, Postmates, Lyft) to get you to use the service again. Blue Apron, same:

FB ad served to me. The $30 meal off might not count as marketing expense, but hidden at the gross revenue level. FB ad itself counts as marketing for sure.

It’s hard to know for sure to what extent the omission of “Gross Revenue” hides additional marketing / promotional costs. It’s super weird, I don’t think I’ve ever seen “promotional discounts” hidden at the Gross Revenue line-item in my 5+ years working in M&A as an i-banker and in corporate development.

What’s their churn rate?

Churn is one of the most important KPIs for any service-based company, and I was surprised that it wasn’t included. The only indication of retention is a line about “repeat order rate”:

During 2016, 92% of our net revenue was generated from Repeat Orders, ranging from 89% to 93% across each of the four quarters of 2016. For the three months ended March 31, 2017, 92% of our net revenue was generated from Repeat Orders.

So — 92% is obviously high. But I don’t like this number, and would much prefer a simple monthly churn rate, or a cohort-level retention curve, because doing it at the revenue level makes it easy to manipulate.

This means marketing can be wielded as a more significant lever to get that somewhat arbitrary KPI up. For example, if I received 3 free meal deliveries via a friend’s referral, the latter two meals would still be 100% booked as “net revenue.” Voila, a 66% net revenue rate from repeat orders! And knowing that gross revenue includes promotional discounts (likely weighted towards the first month of customer acquisition) makes the “recurring” net revenue metric that much more suspect.

Just show us a retention curve, please!

Thankfully, this sentiment was shared by a marketing professor from Emory, who took the time to write-up a fantastic analysis on Blue Apron’s churn. The main takeaway is the retention curve he modeled out:

The author uses the modeled retention to then estimate monthly CAC. He also uses Q1'17 gross profit per customer of ~$25 to conclude:

I estimate CAC in Q1 2017 to be $169, which would imply a break-even point on newly acquired customers of ~8 months, implying that Blue Apron will lose money on 66% of its customers.


Interestingly, Blue Apron recently amended their S1 to include 6-month contribution margin averaged across their ‘14-’17 period: $115. Taking their disclosed 6 month revenue over the same period ($410), and assuming 30% gross margins, implies < 6 month initial customer life.

As a final retention estimate / sanity check, a VC at Shasta Ventures (investor in other subscription companies such as Dollar Shave Club, Imperfect Produce, etc.) provided a 3rd party retention study. Here’s what it looks like, with a steeper curve and lower asymptote than the earlier curve:

Based on a representative % of US households. Not actual customer retention, was from a representative panel via analytics company Second Measure

Steep drop-off in the early months. Perhaps that’s because this 3rd party polling includes real-life customers from promotional discounts? I’ll note here (was also noted in the VC’s post) that there is a big difference between Blue Apron’s business, which can have multiple transactions in one month, versus Netflix or DSC which are once a month transactions.


These questions around churn and financial wordplay loom large in my mind, and the fact that Blue Apron deliberately chooses not to address them directly is a bad sign.

To play devil’s advocate, perhaps the key criteria to judge the company is customer reactivation. Maybe high churn doesn’t matter, because customer usage is episodic and not really subscription based. That is, maybe Blue Apron’s business is ok if customers use it for a few months, churn out, then come back months later…etc. However, if this was true, I think Blue Apron would know this and disclose this behavior in their S-1 already.

And even if true, there is still the question of the cost of reactivation, via rising overall marketing costs or potentially obscured “promotional” coupons at the gross revenue level.

Show your support

Clapping shows how much you appreciated Jason Lee’s story.