My views on the market, tech, and everything else

Peloton: The “SaaS Plus a Box” Business case


“We’re a software company. The entire leadership team comes from consumer Internet… what differentiates us is the software, which includes the streaming and the gamification and the network. We’re also a media company on top of that, because we’re streaming 12 hours of live TV content each day and have another 4,000 classes on-demand.”  Peloton CEO John Foley

Peloton sells stylish stationary bicycles and treadmills to customers:


But despite the functionality and style of the hardware the Peloton CEO said in the quote just above that they are a software company. Why is that? Because it is in software that a business can find the best sources of sustainable competitive advantage (AKA, a moat). If a business does not have a sustainable competitive advantage alternative sources of supply will arrive in a market and price will drop to the opportunity cost of capital. Occasionally you will see or hear a nutcase argue that moats do not matter which the equivalent of saying alternative sources of supply of what you sell created by competitors do not matter. Bill Gates described the importance of a sustainable competitive advantage in an interview in 1992: “Supply is the killer of value. That’s why the computer industry is such a strange industry.  We’re dealing with amazing increases in supply.” What any business must find is some degree of scarcity which is what generates a positive return on capital. The best source of that scarcity is software, which today often takes the form of “software as a service” (SaaS).  Some people like to say that “every business is now a technology business.” The more precisely accurate statement is that specifically “every business is a software/SaaS business.”

In his famous 1994 interview in Playboy magazine Bill Gates described why moats are so important in this way by commenting on a specific example: “What is the scarce resource? What is it that limits being able to get value out of that infinite computing power? Software.” Software/SaaS has increasingly assumed the top dog role in creating significant profitability, especially in the start-up world because it is network effects (demand side economies of scale) that create the most meaningful barriers to entry in today’s economy. If all you make is clever hardware the story is almost always that you end up in a situation like GoPro finds itself in today. GoPro is in a business that is increasingly a commodity since it lacks network effects to preserve the source of its margins. Just trying to compete on operational effectiveness alone is a very bad idea. Operational effectiveness is essential but in the long term it is not enough to guarantee success,

The Peloton founder understands the power of software/SaaS to drive return on investment and not surprisingly is focused on that part of his business. Marc Andreessen made the same point in his well-known “software is eating the world” essay. You may be asking at this point: “Why is Peloton making hardware at all if software/SaaS is so valuable? The one-word answer to this question is: distribution. No one is making low cost exercise bike or treadmill “clones” or “blanks” that Peloton can rely on. Bill Gurley described how hardware in some cases can be important for distribution in his 2003 essay: “Software in a Box” which today might be called “SaaS plus a Box”:

“Most executives in Silicon Valley take it for granted that selling software is a better business model than selling hardware. In their minds, this goes without saying. The self-evident reasons relate to software’s remarkably high gross margins. With near-zero variable costs, software businesses offer the ultimate in scalability. Software businesses are simultaneously less capital intensive than hardware. This combination of low capital intensity and high gross margins, also leads to better valuations in the public marketplace. What’s not to like? Despite these seemingly irrefutable advantages, many startups are “choosing” to sell hardware instead of software.”

Gurley wrote this essay just after a period when I had spent a lot of time working in the Benchmark Offices on Sand Hill Road since the private equity firm I worked for then was co-investing with Benchmark when it came to mobility investments. We talked about the “software in a box” idea often then. I was focused on this investing thesis in no small part since: (1) I loved software profit margins especially given some of the other business we were involved in that had modest or low margins and (2) I saw how hard it was to get distribution for a new product offering. Hunter Walk recently wrote about why I love the software business so much:

“Nothing gets an investor’s heart racing like the phrase ‘software margins.’ It’s shorthand for the concept than businesses which are primarily bits (not atoms) have some very attractive characteristics: fixed development costs, economies of scale in deployment & servicing, and “winner take most” markets with pricing power. The resulting impact is very profitable, fast-growing success stories with high gross and net margins. Dismissing a company as ‘nice but doesn’t have software margins’ is the “yeah, nice personality I guess’ of venture investing.”

What are some examples of “SaaS plus a box” today? Peloton? Apple watches? GE Jet engines?  John Deere tractors? Amazon Echo? Kindle? Video game consoles? Fitbit? Roku? The business model is not easy but there are not that many opportunities to create a new platform business so using hardware for distribution can be quite tempting or even necessary. Steve Jobs said in 2004:

“The more we look at it, for more and more consumer devices the core technology in them is going to be software. More and more they look like software in a box. And a lot of traditional consumer-electronics companies haven’t grokked [fully understood] software.”

Hardware may be a valuable way to distribute software/SaaS, but it makes for a much harder business model to execute on. The skills and business models necessary to get the product and service integration right are sometimes conflicting. Getting the hardware designed and manufactured correctly is non-trivially hard. Managing a global supply chain is tricky, as is customer service. The upfront cost of creating, manufacturing and the financing hardware for consumers can result in a higher customer acquisition cost, but the relationship can become quite sticky. But as anyone who has tried to sell hardware devices knows the amount cash required to do so successfully is huge. Up front expenses like non-recurring engineering and work in process make it hard to finance. Add upfront costs of SaaS customer acquisition to that and you are talking about significant cash requirements. In other words, the unit economics of a SaaS plus a box business model can require that the business raise a lot of cash before it flows enough cash to be self-supporting.

As an example of what I talked about above, The Street.com wrote this about how hardware impacts the unit economics of Roku:

“The Roku Player, however, is a true loss leader. Revenue was actually down 7% from a year ago while the cost of that revenue was only down 1%. The company acknowledges the Player as a loss leader, but the concerns become how quickly will it slow and how much will it drag the bottom line. Gross margins for the Roku Player dropped from 14.3% a year ago to 9.5% in the fourth quarter. Fortunately, the Platform gross margins are strong, at 74.6%, but those are also lower by almost 300 basis points from a year earlier.

SaaS plus a Box can work financially, but it is not an easy business model to get right.

Let’s return to talking about Peloton. Where does Peloton’s moat come from?

“When I started Peloton with my cofounders, I saw clear as day what it was going to look like and how it was going to work — the technology, the hardware, the software, the business model. I saw everything except the community. The community has blown me away.”  Peloton CEO John Foley

What a community does, if done right, is deliver network effects. My essay on network effects-based moats is cited in the notes and it reads in part: ““Nothing scales as well as a software business, and nothing creates a moat for that business more effectively than network effects. Network effects exist when the “value” of a format or system depends on the number of users.” Building a community to create those network effects is neither easy or simple. Solving the “chicken and egg” problem is hard. Somehow the services must become viral to drive community growth. Many businesses try to create community, but few actually do it well.

Gizmodo describes some of the key financial terms in the case we are looking at in this blog post:

“Peloton will sell you the new Tread for $3,995, and you can reserve a unit now with a $250 deposit. If you don’t have that kind of coin lying around, you can sign up for a payment plan that costs $149 a month for 39 months—which is, by the way—with 0 percent APR. That payment includes the $39 a month membership fee for the duration of the payment plan. Still, we’re talking $150 a month a little over three years to get the Peloton Tread and the classes. After that, you still have to pay that membership fee.”

Five factors determine the “unit economics” of a business. Let’s work though a key sensitivity analysis using educated guesses as inputs. I will assume that the Peloton equipment business is financially break even to make the math simple. This is an important assumption and the model would be a lot different if the hardware is actually a loss leader.  It could be that the equipment is sold at a loss and if so that is properly classified as being part of CAC.

  1. Average revenue per user (ARPU) per month – $39- This is what we know best about Peloton.
  2. Gross margin – 70%. In a SaaS business the gross margins should be very high. I would have put it at 80% if not for the cost of making the videos and instructors.
  3. Customer acquisition cost (CAC) – $500. Yes, I think it is that high and it may in fact be higher. A business can spend what may seem like a lot of money on customer acquisition and still have at attractive business if the unit economics are otherwise acceptable and the business model is sound. For example, sometimes doing things like this very expensive “truck roll” (it will cost at least $150 for each one) becomes possible if gross margin and revenue per month are high and churn is low.I formulated my $500 CAC estimate by looking at ARPU relative to comparable business, noticing the frequency of their sales pitches on expensive advertising like TV and the way they distribute goods. Peloton’s network of stores is not a cheap way to sell goods either. But again, spending what may seem like a lot on CAC can make sense if the business has (1) solid unit economics otherwise and (2) enough cash to get to free cash flow positive. Peloton is similar to mobile phone customer style ARPU. It is also like a satellite TV install where the product is sometimes brought to the customer via a truck roll.

4. Monthly customer churn – 1%. The company has said “Monthly subscriber churn is just 0.3%” but that will not last forever as the equipment and people age. A 1% churn rate is really good for a consumer product. The best cell phone providers are below 1% a month on churn but not by much. I asked Professor Dan McCarthy for his view on what churn might be for the service. He thinks their retention is excellent, but very uncertain:

“They say their current monthly churn rate is 0.3%, which annualizes in the very worst case scenario (i.e, assuming all customers share the same loyalty) to an annual retention rate of 96.5%. But their true underlying retention will almost surely be higher than that because of their 39 month financing plan, which forces people to be on subscription for 39 months. A lot of their customer acquisition has been more recent, which means most of their customers probably have their “asses glued to their seats” (no pun intended). I don’t know how much higher their normalized base churn rate is, but to their credit, it is probably not very high because they have great user engagement and customers seem by and large very satisfied with the service (their NPS of 93 is impressive).”

Of course, the monthly retention rate of an “average” customer is just one of the two main drivers of expected customer lifetime, the other component being how much variability or “heterogeneity” there is in retention across customers. The more heterogeneous retention rates are across customers, the higher the expected lifetime an average customer has. I have no idea how heterogeneous retention is. However, I would have to imagine there are two segments – the die-hards, and those who are into it but not quite as loyal.

Running a sensitivity analysis across possible base retention rate and heterogeneity assumptions, I would estimate an average customer lifetime somewhere between 7 and 14 years. While this is a wide range, these values are high, which Peloton should be proud of. Of course, Peloton would have a much better sense for their expected customer lifetime because they have access to their own internal transaction log data.”

  1. Discount rate – 10%. This assumption seems about right but you can select another discount rate for your model if you want.

How big is Peloton now in terms of revenue? In a Facebook post, the CEO shared this chart:

Peloton LTV

Peloton sells hardware and SaaS, so the revenue is higher than a software or SaaS business, but still that screenshot of the spreadsheet above chart reflects significant revenue growth. It should be noted that Peloton generates a large amount of customer data that could be very valuable to e-commerce firms and that could be an additional source of revenue. One uncertainty about the Peloton business is the size of their total addressable market (TAM).  How many people are willing to pay this much for equipment and $39 a month for service? Estimating TAM is a topic for another post.

It will be fun to watch how Peloton executes on their business plan.






Gurley: http://abovethecrowd.com/2003/02/10/software-in-a-box-the-comeback-of-the-hardware-based-business-model/

Andreessen: http://online.wsj.com/news/articles/SB10001424053111903480904576512250915629460

Gozmodo https://gizmodo.com/pelotons-new-treadmill-costs-4-000-and-oh-crap-i-want-1821952848

Peloton revenue: https://www.linkedin.com/pulse/bigger-than-ring-why-acquiring-peleton-hot-iot-startup-jon-nordmark

Hunter Walk: https://hunterwalk.com/2018/03/02/is-the-quest-for-software-margins-to-blame-for-twitters-trolls-facebooks-russians-and-youtubes-fake-news/

Roku:  https://realmoney.thestreet.com/articles/02/22/2018/roku-how-play-and-think-about-battered-stock-after-its-latest-results

Network Effects: https://25iq.com/2016/03/24/two-powerful-mental-models-network-effects-and-critical-mass/



Categories: Uncategorized