So, I know the WeWork S-1 just dropped, and if you’ve been following me on Twitter any time over the past ten years, you probably know where I stand and what I’m going to say about that nonsense. But first, I want to talk about Tumblr, who Yahoo!/Verizon just sold to Automattic for $3 million which is, uh, quite a bit away from the $1.1 billion they paid for it.
Now, to get it out of the way: I liked, but didn’t love, Tumblr. As you might be able to tell from my infrequency of posting here, I’m not much of a sharer. My lack of personal utility for it doesn’t mean however that I didn’t appreciate it for it was: a true evolution of blog software. Just as Tinder begat OkCupid begat Match begat eHarmony, as verticals matured and as mobile phones became more capable, the products designed to serve those markets got smaller and simpler. And so Tumblr became the defacto software of choice for a whole new and younger generation of users, for which WordPress (ironically, now it’s corporate big sister) and Squarespace didn’t provide much utility, or worse, was a downright grandmotherly way of doing things.
Of course, that new and younger generation was exactly the problem. But instead of analyzing that directly, I want to get to my eventual point by asking an open question that I don’t think anyone has really answered to any level of satisfaction: why the hell did Yahoo! buy this thing?
Let’s start with some assumptions which are almost definitely true but I don’t know for sure because I don’t work at either company involved:
- Due to the disruptive nature of the platform, the user base was materially younger than competitors and thus, disproportionately low-HHI, if they had access to wallet spend at all
- Due to the openness of the platform and lax moderation which tacitly encouraged acceptance of any content, a large chunk of the marketable inventory was not work- or brand-safe
- Because users weren’t tied to real-world identities and because the quality of tagging and other ontological meta-data was low-quality at best (certainly compared to say, Facebook), accurate targeting on any reasonable criteria (location, interest, gender) was impossible
Sounds like a lot to overcome, right? I agree! But remember that a suit somewhere in the deep recesses of Sunnyvale decided it was worth $1.1 billion, so let’s now dig into what that person would have had to believe to justify the cost.
- The brand would start to attract an older, more valuable demographic, despite being designed in a way that is experientially more difficult and not tailored to them (see also: Snapchat)
- Users would not leave en masse once brand-safety initiatives were put into place to allow advertisers comfort that their brand would not be adjacent to objectionable content (see also: Snapchat, Reddit)
- Some other business model beyond chum remnant display inventory would emerge, despite that not happening in any large-scale consumer product that I can readily name (see also: Snapchat, Reddit)
- If #3 doesn’t happen, then a reverse of the trend of display advertising CPMs bottoming-out asymptotic to zero because users are blind to them and because Google/Facebook have strangled the market
If any one of those four actually happened, Tumblr would be thriving right now. #1 and #2 would have gotten Tumblr to a scale that not even the worst operator could have screwed up, #3 would have been a severe long-shot but would have created a viable LTV even if #1 or #2 destroyed actives, and #4, well, that’s just wishful thinking on multiple fronts but hey, the other three were pretty much pipe dreams too, so why not?
Too bad hope isn’t a particularly strong strategy!
The common maxim is that Yahoo! bought Tumblr basically because it used to be cool and then all of a sudden, it wasn’t, and we all know in tech that once that coolness factor is gone, the downward spiral is nearly impossible to pull out of. Well, sure, I get that. And it’s true – Yahoo! definitely wasn’t cool at the time; it had badly missed social and mobile, and hadn’t been relevant in the market for top technical talent in about fifteen years. But that wouldn’t have been sound logic even if they didn’t have a history of screwing acquisitions up (see also: Flickr, Polyvore) and ham-fistedly integrating them with their outdated technologies.
We know what happened: the brand never grew beyond a relatively niche community, the ban on sexually explicit content cut the legs off a vibrant (and underserved!) segment of their userbase, and even despite that move towards brand safety, the CPMs weren’t enough to make it worthwhile, particularly not with Yahoo!’s substandard ad-serving technologies and the complete lack of useful targeting and segmenting.
And then Automattic bought it for $3 million, and we move on to the second part of the story.
On the many Hacker News posts about the transaction, I saw many variations of the same post, all generally aghast at the low price Automattic paid. Here’s an example:
Considering the fact that Tumblr still does over 2.5 billion page views per month, not even counting their mobile usage, I can’t help but gawk at what a steal Automattic got. Really drives home what their CEO Matt Mullenweg said on here the other day about adopting a “Berkshire model”—although he was referring to independent management, what he’s really talking about is buying companies grossly undervalued for dirt-cheap. Just monetizing the web traffic 2.5 billions monthly hits at a $1 CPM alone would generate $2.5 million a month in revenue, $30 million annually. What kind of business, let alone web business, sells for 1/10th annual revenue? For reference, Reddit does 6x the page views, but is valued at $3 billion.
(For the sake of brevity, I’m going to assume that the page views estimate is valid. But let’s break the other inferences down.)
First, let’s look at the $1 CPM. Average CPM rates on the big four social media sites (Facebook, Twitter, Instagram, and LinkedIn) all average around $7, so at first glance, the estimate here seems cheap. However, bear in mind that Facebook has bar none the best targeting in the game, and Instagram is basically the same thing. LinkedIn has real-name attribution and is massively brand-safe, given that it’s a professional network. And while Twitter doesn’t have the last two, it does have a massive trove of semantic information about a user in the form of their text-based UGC in order to target against. Tumblr has none of those things – nor the coolness factor nor scale as the others – but for the sake of argument, sure, let’s stick with $1.
So, $30mm annual revenue. OK, so best case scenario, what, Yahoo! bought this thing for 37x revenue? Well, that was dumb, but Automattic was smart, they bought it for 0.1x revenue. What a move!
Well, not really. The $30mm figure is revenue, not discounted free cash flow. (I’ll touch on this a little bit later, but the distinction between revenue and cash flow has never been more obscure or meaningless in Silicon Valley. But that’s a whole other topic.) Since running a website requires servers, employees, Patagonia vests and a whole mess of other stuff, it’s likely that even at this inflated revenue estimate, Tumblr was losing money. How much money? Let’s try to ballpark that too, because I’m on a flight and I’ve got nothing better to do!
It’s been reported that as a condition of the sale, Automattic agreed to onboard and hire the 200 employees currently working on Tumblr. (Before you consider that a grand gesture, assume that it would be worthless without them.) Considering that we’re talking about the Bay Area here and considering we’re talking a company that was likely tech-heavy just due to the nature of the work, I’ll average salaries out to $150,000 and add on 50% cost to the business for benefits, insurance, taxes, and the other cherries on top. So, round numbers: $45,000,000 in operating cost just from labor alone, or 1.5x an already confident estimate of revenue.
In good times, be it an exogenous bull market or a hypergrowth phase for a company, the acquirer is very often willing to pay exorbitant multiples on revenue. This may be for a variety of reasons: maybe the company thinks it can operate it more effectively, maybe there is IP that is accretive to the acquirer’s operations, or maybe it wants to take a competitor off the market. Or maybe, just maybe, it’s just not the right time for this company to focus on making a profit. This is an area in which a lot of people who sit outside of the tech sector (particularly the venture-backed tech sector) have a hard time understanding the voodoo economics at play. Even though most people who follow me know this already, there’s an outside chance my family will read this (hi Joanie!) and want to know more, so I’ll do my best to simplify it:
Venture capital exists as a way for limited partners (ie: those who invest in the VC firms themselves) to take their capital and allocate it to high-risk, high-reward investments. This is often a diversification and upside exercise, as a majority of their assets invested elsewhere in low-risk vehicles. In order to fulfill the mandate of high-reward outcomes, the VCs in turn only make investments that they believe have massive potential. Some do this as a function of the total addressable market, some do this as a function of trend prognostication, and there are loads of other strategies too – but in the end, the goal is the same: hit home runs.
Now, to hit a home run, growth is often much more important than short-term profitability, and this is where the significant deviation from almost every other business occurs. In many markets, there is a winner-take-all effect (or in a new one, a first-mover advantage) therefore carving out as much space as possible is a winning strategy, and often impossible without the help of venture capital.
Of course, venture-backed growth can sometimes obfuscate huge and glaring holes in the fundamentals of a business. (Don’t get me started on Uber or Postmates or really the entire on-demand sector, which seems to have forgotten how unit economics work.) Although we’ve sacrificed short-term profitability in the name of strategic growth, we have to eventually be able to stabilize as availability for venture capital is not infinite, nor is it immune to macroeconomic trends going on elsewhere. Using Tumblr as the example here, much like Facebook and others, it was assumed on the basis of the user growth that it had the potential to grow to such a scale that even a beaten-down business model like display advertising could grow to be a huge, profitable enterprise. In hindsight, it’s easy to laugh but at the time in which Tumblr was receiving this venture capital, the outcome was unknown and indicators looked reasonable to investors who had a certain thesis and appetite for taking on that risk.
Like a lot of things, however, the devil is in the details. Investing in Tumblr in the really early stages is different than doing so in the later stages when some warts are visible, and it’s really different than buying the company when the warts have turned into something that you’d have to be incompetent (or Yahoo!’s M&A team) to ignore. There’s a whole lot to unpack there, too: sunk-cost fallacies, corporate strategy…but this thing is already like 3,000 words and the flight is almost over.
So what do we conclude? Up to you, but here’s mine: Yahoo! made a ridiculous acquisition using assumptions that don’t make any sense, Automattic made a slightly more reasonable acquisition but the way most people are talking about is misleading at best, and the tech/VC sector writ large still has a very difficult time figuring out how to effectively assign valuation and how to monetize audience.
Anyway, thanks for reading. I’ll try to write more often. And really, that WeWork S-1; if there’s anything that will get me to back to the world of writing, it’s that mess. Lots of hot opinions coming soon about that, I’m sure.
My thanks to Will for usually being the only person to read this stuff in full before it gets posted, and thanks to the patient team members on the numberFire Slack who have heard and are likely sick of me railing on about this stuff for years and years.