by Will Schreiber

Restaurant Revenue and Goodhart’s Law

When we moved to Ketchum two winters ago, Elizabeth got a job waitressing at a little eight-table local restaurant.

The walls were still blank on her first day of work. They’d only been serving guests for a week.

January 2019 marked the restaurant’s first full month of pumping out pizzas and wine. They went on to do ~3x more “monthly” revenue than we did that same month at Bottle. And we had a three year head start!

If each business was reduced to an Excel sheet comparing only revenue, their “$100k of sales in the first month” narrative would outshine our “5-digit MRR three years in” narrative.

But pizza comes with significant marginal costs: flour, San Marzano’s, mozzarella, cooks, hosts and waiters. There’s a reason I’m happy owning Bottle’s current and future cash flow instead.

Everyone jokes about using revenue as a metric. “What we lose per unit we’ll make up in volume!”

But I’ll bite. For software businesses, revenue is a meaningful metric because costs are fairly fixed and familiar: engineers, ads, and AWS bills. This is why the entire SaaS market has been trading ~10x revenue, independent of vertical.

“Fixed and familiar costs” don’t apply to the newest venture-backed businesses, though. As revenue becomes the ultimate metric, software-enabled companies are hacking top line revenue. It’s Goodhart’s Law.

I saw a tweet by Garry Tan yesterday, sharing Truepill’s latest $25 million round:

“Truepill revenue reached $48 million in 2018 and nearly $100 million in 2019; it could reach an estimated $200 million this year as it pushes further into working with health plans, drugmakers and pharmacy benefit managers.”1

Maybe those revenue numbers are compelling. Maybe the aren’t. I can’t tell because revenue has lost its meaning. It got me thinking about three things:

1. Restaurant Revenue

I propose a new term: Restaurant Revenue. Restaurant Revenue is revenue that comes with real and significant costs. A pizza. A pizza delivery. A short-term tenant inside a long-term lease. A flu shot. A prescription. These things are inherently different than HTTP packets.

Pharmacies in particular come with incredibly high marginal costs.

I used to think Walgreens used Doritos and Coke to lure people in to buy high-margin drugs. It’s actually the other way around. The pharmacy is a loss-leader to get people to come buy toilet paper and razors. Pharmacies are so low-margin that the undisputed king of low-margin is struggling to make drugs profitable. Their rival Target gave up completely.

Truepill is an API-enabled pharmacy. It’s a great idea. But how much of the $200 million in projected revenue is zero-margin prescription drug revenue?2

If it’s mostly Restaurant Revenue, then it’s a vanity metric.

2. Alpha Revenue

Restaurant Revenue isn’t all bad. Lots of software-enabled companies genuinely reduce costs and induce demand in novel ways.

Thus I propose a second term: Alpha Revenue. Alpha Revenue is the contribution margin that, with a straight face, you can say is added because of software.

Cloud kitchens and Chick-fil-a both sell fried chicken sandwiches. Let’s say a cloud kitchen has trained a robot to hand-batter all the chicken breasts. In doing so, they might knock $0.50 in cost off each chicken sandwich. Alpha Revenue.

Or take Bungalow. They can induce demand from a well-designed website. They can make it easy to split the house and rent by the room. This enables them to turn 3-bedroom houses renting for $3k/mo into 4-bedroom houses renting for $1.2k/room. The Restaurant Revenue is $4.8k/mo, but the Alpha Revenue is $1.8k/mo.

3. Business Models

Eventually, the market will sort the valuations out. But, for any business that lists software engineers in its S1, the market is struggling to discriminate between Restaurant Revenue and Alpha Revenue.

It’s how you end up with an app-driven insurance provider worth $4 billion.

The downstream valuations are having a negative impact on new startups. There’s a pressure to build revenue as fast as possible, Restaurant Revenue or Alpha Revenue be damned.

My friend is starting a women’s hormonal health company. She wants to pair women with PCOS with providers. VC’s are pushing her to “own the entire stack” and bring healthcare providers onto payroll. It’s an easy way to book a lot of top line Restaurant Revenue. She’s wary.

In fact, Truepill just invested in Ahead, a company with full-time therapists on staff ready to provide Telehealth treatment for ADHD, depression, and anxiety. Lots of Restaurant Revenue. Can they produce Alpha Revenue, though?

If Airbnb was raising a Series A today, it’s tough to imagine VC’s not pushing them to “buy the real estate yourselves.” Get the top line revenue to swell, raise the next round, pass off to the public market.

There are better business models than restaurants. Google, Lyft, and Airbnb all invented not only new products, but new business models. Where did all the business model innovation go? I don’t know if it’s a failure of vision or the fact that VC’s got sick of battling independent contractor employment law.

I’ll be curious to see which of these restaurant models end up with enduring Alpha Revenue.


  1. Truepill revenue reached $48 million in 2018 and nearly $100 million in 2019; it could reach an estimated $200 million this year as it pushes further into working with health plans, drugmakers and pharmacy benefit managers.

    — Garry Tan (@garrytan) July 8, 2020

  2. Based on known public information, I think Truepill is a fairly capital efficient business. My criticism of revenue as a metric is not a criticism of Truepill as a business.