Warby Parker and EssilorLuxottica: Irresistible Force, Immovable Object
Part of the story of many successful companies is that 1) they find an almost magical growth strategy that allows them to operate with completely different economics from their competitors, and 2) step by step, they copy more and more of what those competitors do. It happens everywhere: Amazon killed department stores and is now opening department stores; Netflix used an asset-light model that relied on other people's content and now depends on its own content; Zillow built a valuable business by essentially earning a low-risk royalty on home sales, and then pivoted into the much higher-risk business of buying and selling homes itself. And Warby Parker started with a bold question: what if we could buy glasses—online? And, three years later, it was asking the equally bold question: what if you could buy Warby Parker glasses—in a store?
What's especially interesting about Warby Parker, whose S-1 dropped on Tuesday, is that they're a growing business in an industry that is, at least according to some observers, under the control of a powerful, grasping monopoly. EssilorLuxottica is a French/Italian eyewear conglomerate that has been snapping up low- and high-end brands, from fashionable sunglasses to prescription glasses, and now owns or licenses an impressively long list: Oakley, Dolce&Gabbana, Oliver Peoples, Prada, Ray-Ban, Valentino, Versace, and many more. If there's a luxury eyewear brand, there's a good chance EssilorLuxottica owns it; if there's a general luxury brand, there's a good chance they're the ones who make the glasses. They're also likely to own the stores, especially since they offered to acquire a 7,000-store chain in 2019, in a deal that closed earlier this year.
The eyewear business is a good place for a monopoly to operate, since there are so many opportunities for price discrimination. Glasses can be a healthcare expense, where no one likes to economize and many customers aren't spending their own money.1 And they're also fashion, an area where buying the most ridiculously expensive brand is an easy heuristic. It's unclear exactly how much market power EssilorLuxoticca exercises, and how well this can be differentiated from the usual margins that a vertically-integrated seller of luxury brand products would achieve. But they certainly get accused of it, at length.
So it's interesting to look at how Warby Parker has grown in that environment, and what they do differently.
Warby Parker's fundamental idea is to 1) make the glasses purchasing process easier, both in terms of convenience and price, and then 2) extend that ease-of-purchase through as many channels as possible—a website, an app, and physical stores. Notably easier purchases do not just have a linear effect on the economics; they make it an entirely different business. Glasses are a repeat purchase, since eyes adapt over time. Warby cites a 50% sales retention rate over a two-year period and 97-98% rate over a four-year period. In other words, a one-time sale for $100 is really a sort of annuity paying about $24.75 per year thereafter. There are incremental costs here—their contribution margin from incremental sales is 20%—but some of those costs are things like store rent and ad spend, which could go down over time as more of their customer base becomes repeat customers. Meanwhile, customer count has compounded at 25% annualized since the beginning of 2017.
In a business with repeat transactions but not subscriptions, the company that optimizes for satisfactory unit economics and then maximizes its net promoter score and the amount of data it collects from customers is likely to win. Warby Parker lists its high net promoter score of 83 prominently in the S-1; it's the third metric they highlight in their summary, after revenue and revenue growth.2
Customer recommendations are useful, but a company that wins in only one channel is wasting them: someone who recommends Warby Parker to an ecomm-unfriendly friend is probably not going to convert that friend to both Warby Parker's brand and the concept of online shopping in one go. So the company aims to be channel-agnostic; they're indifferent to where they sell glasses, as long as they successfully sell them. Pre-Covid, their sales were 65% retail stores and 35% online, though in the last six months the number was 50/50.
This ties in nicely with the data-heavy approach: a company like EssilorLuxoticca is partially vertically-integrated, since it owns thousands of stores. But it also sells its products through many more independent stores, and since it makes lenses and frames, is also sometimes selling a component rather than a final good. At every stage of the supply chain that separates them from their end customers, they're bleeding bits of data that could help them resell to those customers.
A pure retailer can do plenty of cohort analyses of customers, but is limited in what it can do with that information; it can choose different inventory, and adjust its marketing, but it can't easily go where the market is going. A pure manufacturer has lots of control over the product, but less information about individual customers; its retention economics are at the level of retailers, not customers. Warby Parker chose to be vertically integrated from the point of sale up through product design, but, interestingly enough, not at the level of manufacturing: they still buy products from external suppliers, but say that their top five suppliers are only 23% of sales. This implies plenty of fragmentation among manufacturers.
Warby Parker is a young brand but a bet on aging: as they note in the S-1, 84% of people over 65 need some form of eyewear, and that population is the fastest-growing in the US. There's a whole category of companies that owns its main demographic but is stuck on a treadmill of constantly replacing them; Snapchat, TikTok, Roblox, and Nickelodeon all have to convince the next crop of 18-, 15-, 10-, and 6-year-olds that they're worth paying attention to, and each one's target demographic churns out. For Warby Parker, the churn dynamics are reversed: the thirtysomething customers who like them will age into being the fiftysomething customers who need them, and who have a lot more disposable income besides.
So Warby Parker ends up being a meta bet on eyewear: it's a bet that the winner is the company that expects to keep its customers the longest, and builds its distribution around knowing who those customers are so it can measure just how likely they are to stick around. The institutions that do this tend to win over time (as I noted in a very different context earlier this week ($)) and it's hard to tease out causation. Does Warby Parker have a model that assumes customers will be happy to keep coming back because it believes its products are better, or did it choose that model specifically so it could get a higher incremental return on its stores and ad spend than the incumbent? Either way, when comparing two competitors with different time horizons, the default bet is that if the one with the longer time horizon can survive, it's the one that will win.
A Word From Our Sponsors
Secureframe helps hundreds of companies get enterprise ready by streamlining SOC 2 compliance. Secureframe allows companies to get compliant within weeks, rather than months and monitors 40+ services, including AWS, GCP, and Azure. Secureframe continuously collects audit evidence, runs security awareness training, manages vendors, monitors infrastructure, and more, all automatically.
Our team of compliance experts and auditors are happy to help answer any questions and give you an overview of SOC 2, even if you don't need it today!
Our customers save an average of 50% on their audit costs and hundreds of hours of their time. Schedule a demo and learn how!
Cutting the Apple Tax
Apple made a surprise announcement that they're letting developers inform users of ways to pay for their apps outside the app store, conceding a point that they've fought over for years. This is especially important for any app that has a fixed cost per user; paying 30% to Apple can mean the difference between earning a decent profit and operating at a loss. Apple has also committed to making the app store ranking algorithm focus on objective criteria, though one of those criteria, "user behavior signals," still gives them ample wiggle room.
One side effect of this change is that it will make companies more willing to market their app, since the app tax is no longer a mandatory component of their cost structure, so Apple is trading some revenue now for more developer lock-in over time. It will be especially interesting to see how casual game companies change their behavior: a big driver of their revenue is how easy it is for players to casually spend money, but given how much spending is skewed to the small population of "whales," it may make sense for those game companies to send such players special offers outside of the app store. And that has a direct effect on Facebook, since casual games set the reserve price for Facebook ad inventory.
This also gives Apple an even bigger incentive to sell its own apps. If the iOS ecosystem is stickier to developers, it will be stickier to users, too, so Apple can expect (and monetize) even more compulsive iPhone use than before. So, as with many other Apple announcements, it's both good news and bad news for developers. The good news is that they have an opportunity to take back thirty points of margin on transactions, and the bad news is that they have to worry about a big, well-funded competitor in Cupertino.
It's hard to summarize Google's struggles with messaging, but a good short way to do that would be "It took 25,000 words to detail all of Google's efforts to build a messaging product." You can read those words at Ars Technica.
There are two kinds of companies that succeed at messaging: the ones that have bet their company on it because it's their only product, or the ones that bet the company on it because it's an existential risk. Facebook knows that monthly active users are a lagging indicator of daily sessions per user, and has acted accordingly by building a solid and tightly-integrated messaging product and buying the most threatening competitor. Meanwhile, Snap is messaging, and will never risk losing that. Paradoxically, the problem Google runs into in messaging is that losing that market is not an existential risk; messaging doesn't give great signals for search (although it might surface a handful of obscure URLs and reinforce the interest graph a bit). And Google doesn't have to worry about logins because Gmail and YouTube are so ubiquitous. So messaging ends up being a nice addition to many Google products, but never a market the company can't afford to lose.
China and Security
Politico highlights the Chinese government's growing focus on cloud computing, a market that is obviously of interest to a) a country whose economic planners want to bring some demand for its exports back into the country in order to be less beholden to everyone else's economic fluctuations, and b) a party that is much more comfortable when it has control over citizens' data. Meanwhile, state-sponsored hackers are increasingly being outsourced to private sector or nominally private sector firms. This gives the state more flexibility, but at a cost:
“The upside is they can cover more targets, spur competition. The downside is the level of control,” said Robert Potter, the head of Internet 2.0, an Australian cybersecurity firm. “I’ve seen them do some really boneheaded things, like try and steal $70,000 during an espionage op.”
This point is mostly worth bringing up because it's a good example of science fiction accurately predicting the future. Hackers with seriously evil intentions who get caught because they engage in some minor theft along the way sometimes show up in stories because they're good for the plot (it's a great way to escalate the implausibility of a story over time), but it's also a meaningful risk for cyber attackers: when they outsource, they're working with someone who has demonstrated both technical aptitude and willingness to break the rules, and it's risky to assume that there are some rules they won't break.
Cyber Risk and Coordination
On Wednesday there was a spate of security-related announcements from big tech companies, prompted by a meeting between President Biden and the heads of several major technology companies. One reason cybersecurity has gotten less investment than it should is the externality problem: it's one thing for Apple, Microsoft, and Amazon to secure their own products, and another matter entirely for them to ensure that everything built on their platforms is safe—that requires some pretty high-level thinking, since they have to model both what the smartest attacker would do and what the least smart victim would do. One way to solve externalities, though, is to get many interested parties to invest in them all at once, and that seems to be the purpose of the meeting.
Org charts can be functional (everyone doing finance reports up to the CFO; everyone in marketing up to the CMO) or divisional (everyone in the Widget Department reports to the head of the widget department, whether they're doing finance, marketing, or something else). And executive training can work the same way: one model is to give someone steadily escalating responsibilities until they're handling almost as much as the CEO, and then make them CEO, but the other option is to have them work directly with the CEO. Amazon has deliberately incorporated the latter model (as did Jeff Bezos' prior employer, D.E. Shaw); the technical advisor to the CEO essentially solves every problem that a) has made it all the way to the CEO, but b) that the CEO doesn't have time to solve just yet. (For a look at what this is like, Working Backwards, was coauthored by someone who had that role.) Wei Gao, who recently held the position, has been hired as COO of Hopin (Edit: COO, not CEO), a virtual events platform with $1bn in funding. There are other anecdotal cases where someone who shadows a CEO ends up working as one later on, but it seems like a model that could be applied more. Economic growth is partly limited by the supply of truly excellent executives, and it's a potential shortcut to finding more of them.
In its S-1, Warby Parker discloses some regulation-induced seasonality: like many companies, Warby Parker gets more purchases at the end of the year. Unlike retailers, though, it's not because of gift-giving, but because people want to use up insurance benefits by the end of the calendar year. This gives Warby Parker an interesting marketing dilemma: they can capture some demand right after the Christmas spending boom by doing a marketing push very late in the year—but the holiday shopping season during which ads get pricey tends to end a few hours later each year as shipping gets faster and customers adjust to this.
This kind of fine-grained seasonality has sometimes thrown off traders who use high-frequency datasets; if purchases shift later in the quarter, then year-over-year results look worse intra-quarter and suddenly improve in the last week or so of the quarter. Since that's also a popular time for vacations, and a time when liquidity is low, they have occasionally gotten stuck short a retailer that was underperforming in the most important quarter of the year until roughly December 23rd, when it flips to beating expectations. ↩
Net promoter scores have gotten very common in S-1s recently. One reason is the usual two-sided network effect: they used to be a more obscure metric, so they sounded like a vanity metric, but now that everyone knows what they are, more companies have a reason to brag about theirs. There's a subtler reason, though: NPS encourages you to value a company based on future growth, and implies that its present customer base will lead to organic growth over time. So high-NPS companies have longer duration than low-NPS companies, and thus benefit more from low interest rates. ↩