Stratechery is a blog about strategy and technology. It’s written by one of the planet’s best tech writers; Ben Thompson. He has developed a powerful framework for understanding the big internet companies, called Aggregation Theory. The framework explains how these companies leverage the internet to reshape value-chains, gain power and extract profits. It helps explain everything from how Facebook disrupts publishers to why Spotify is not simply a “Netflix for music”.
Unfortunately, grasping the current state of Aggregation Theory requires reading about a dozen fairly long posts. You could read this Stratechery article from 2015 where the term was initially coined, however, the framework has evolved a lot over the years. I often see it applied overly simplified with phrases like “user experience is important”. Therefore I decided to write a primer and overview of the topic.
This post is for anyone that is curious about the intersection of technology and business but never had the time to dig into Aggregation Theory. Note that I have no affiliation with Stratechery (beyond being an avid reader and happy subscriber) and I strongly encourage you to read and subscribe to Ben’s outstanding blog.
On to the Aggregation Theory introduction.
Before the internet
Before the internet, the hardest part of building a consumer-facing business was to actually reach the consumer. For instance, during the last part of the previous century, a lot of people could produce text just like a journalist on their computer. However, printing and distributing that text was very hard. You would need to invest considerable amounts in printing facilities and transportation of the paper. In addition, advertisers paid for eye-balls so there was also a network effect for established newspapers. Because of these barriers to entry, you would usually only find a limited amount of newspapers in a given region — starting a new one was really hard.
There are typically three types of players in any given supply chain: suppliers/producers, distributors and consumers. There are thousands or millions of consumers, while there are only a certain amount of factories, printing presses, trucks and so forth. Therefore, it was more profitable to integrate the supply and distribution within a company and optimise those processes. You could not optimize for every single customer as that would have been way too costly: the marginal cost of serving each consumer would be prohibitively high. Instead, the consumer received what came out in their local store’s shelves. Again, the hardest problem was actually reaching the end-user, providing the distribution.
A newspaper could, for instance, have a hundred writers, a printing press and some trucks that could reach something like a few thousand to a couple of million readers. Obviously, all of these readers got served the exact same content.
After the internet
The internet changed that. While you previously would have about the same media consumption as your neighbor, the internet allowed you to consume all kinds of information from all over the world. This was made possible by the fact that sending a piece of text from a computer to the whole world was now a trivial task. That meant that reaching the consumer with some digital information was suddenly easy (there’s no marginal cost), whether that information was the product itself (as with news) or if it was information about something you can buy (as with a T-shirt).
Interestingly, the consumer can not only read the local newspaper and shop at the local fashion store but browse a practically infinite amount of content sources and clothing outlets online. Before, the access to consumers was constrained by the geographical presence and shelf space (i.e. distribution). When anyone could suddenly reach the consumer, we went from scarcity to abundance of consumer touchpoints.
The new scarcity is now that consumers only have 24 hours a day, and the hardest problem is now actually capturing the end-users’ limited attention. The big question now becomes: where do consumers actually look? The answer is, you guessed it: aggregators.
You have probably already figured out that the internet aggregators include Netflix, Uber, Airbnb, Twitter, Snapchat, Facebook and Google. So, what defines an aggregator?
What defines an Aggregator?
Ben lays out three key characteristics an aggregator must have:
Aggregators have a (1) direct relationship with their users. The aggregator always represents a destination that consumers interact with; you practically never interact with their product through another party. You visit the aggregator’s app or website directly and they, crucially, own the online customer relationship. That is why the theory is called aggregation theory: these companies aggregate the consumer’s demand in one destination. Google for search, Airbnb for beds and so on.
Aggregators have (2) zero marginal costs for serving new users. When you’re aggregating millions or billions of consumers at your destination, it’s crucial that you can serve a new user without incurring costs for it. While a newspaper needs to print more paper and then put it on a truck to serve a new area Facebook has no such costs for serving a new area. This is of course uniquely enabled by the internet, but not all internet-enabled businesses have this property. Amazon is for instance in many ways an aggregator, but absolutely incur a marginal cost when they ship physical goods to your door.
Thirdly, aggregators are (3) demand-driven multi-sided networks with decreasing acquisition costs. This is a bit of a mouthful, but it’s actually not too complicated: There are network effects at play here meaning that the aggregator facilitates interaction between two or more other parties; multi-sided. Demand-driven comes back to characteristic number 1: the network effects are driven by the fact that users go to the aggregator. This, in turn, makes it more appealing for suppliers to be on the destination, which in turn means more users want to browse the bigger supply.
All-in-all this means that it’s easier and easier for the aggregator to get new suppliers and users on to their destination and that is decreasing acquisition costs. This is what Ben calls the virtuous cycle.
A good example is how Google created a great search product that brought users (i.e. demand) there in the first place. Given that plenty of users went to Google, most websites wanted to be easily found on Google (i.e. the supply side) and thus adjusted their websites so that Google could better discover them. This in-turn makes Google’s search product and user experience better, which attracts more users and so the virtuous cycle goes.
For Google, this went so well that an entire industry, namely Search Engine Optimization (aka SEO), spawned to help feed Google’s algorithms. Note that in this process suppliers are also commoditized, meaning that it’s hard to differentiate oneself from competitors and you can’t really sell your product anywhere else. The outcome is typically pressured profit margins for all the suppliers — we’ll get back to this point soon.
Different type of aggregators
Obviously, the various aggregators are quite different businesses; Netflix lets you watch a movie while Facebook lets you interact with friends and Uber helps you get a ride. Only very few companies are “perfect” aggregators so Ben split them into a few different categories:
Level 1: Supply acquisition costs
Netflix falls in this bucket since they actually have to acquire their own supply. So why are they even an aggregator? Because they own the customer relationship, they can serve new customers without any transaction costs and they have a network effect in the sense that the more users they have, the more money they can spend on content and every new piece of content decreased the acquisition costs for all future customers. Produced content is a fixed cost that has value for all future Netflix subscribers. As opposed to the traditional TV industry that can only show 24 hours of TV a day and thus can’t show their entire backlog of content at any given point in time.
A crucial point here is that Netflix pays a fixed fee for its content. This gives them a classic software company type of revenue and cost structure:
Where users and revenue can go to the moon while costs stay as good as fixed. When the unit economics looks like this, one can produce an extremely profitable business if you manage to scale to a large enough customer base.
Now, let’s have a quick look at what happens when marginal costs are not zero. Spotify might not be or become an aggregator because they have marginal costs that they pay to the record labels. When revenues go north the costs also do.
The main reason for this is that the three major music labels own the rights to most of the music that people listen to. The supply side of the music (rights) market was very concentrated when Spotify started out. The concentrated supply-side left Spotify relatively weaker than in industries with a large number of fragmented suppliers.
The biggest aggregators often start out in markets where the supply side is fragmented. It’s typically the long tail of suppliers that start using the aggregator because they are more willing to accept any terms to reach new consumers. First, a large number of smaller suppliers get on board and the virtuous cycle works its magic for a while. Then the more established players come along and bring their supply to the platform.
Level 2: Supply transactions costs
These companies do not bear the burden of having to buy their own supply as Level 1 Aggregators, but they do incur a marginal cost when new supply enters their marketplace. In this category, we find Uber and Airbnb. Uber, for instance, have to do background checks on drivers and adhere to local taxi regulation. This means that supply can not come freely onto their marketplace. Level 2 aggregators usually operate in a regulated environment where safety is a concern. In other words, growing their supply-side always requires some work from inside the company.
Another interesting question looming is if Uber truly manages to become an aggregator. At the core of this question is if they manage to aggregate all the demand and fully commoditize the supply side. Lyft is currently a competitor with significant demand as well, and drivers can easily switch between the two apps. So Uber seems to be lacking the lock-in and winner takes all effects usually seen for other aggregators where supply has nowhere else to go.
Level 3: Zero supply costs
Better yet we have the Level 3 Aggregators which can onboard an arbitrary amount of suppliers without incurring new costs. These companies neither own the supply nor have to pay for getting it onboard. Social media sites like Snapchat and Twitter and the app stores are examples. The former relies on ads while the latter takes a cut of purchases of apps (or inside them). At this level, the supply and demand side of the marketplace can grow in the virtuous cycle without any friction. It does not cost Twitter anything to let a new user create an account or a new Tweet that can be served to millions of users.
The super-aggregators arguably have the best business model of all time. As with level 3 aggregators, their supply and demand sides scale perfectly, but their ad sales also scale perfectly. These are, of course, Facebook and Google: two of the world’s biggest companies that to this day, approximately 20 years after their inception, keep on growing their revenue by twenty-something percent year-over-year.
These companies are so unique because they allow advertisers to advertise with self-serve solutions. In other words, they can perfectly scale supply, demand, and revenue. This sets them apart from Twitter and Snapchat that to a larger extent rely on their sales force to sell ads and grow their revenues. Advertisers on Google and Facebook mostly don’t speak directly with any human while spending money on their platforms.
Now that we’ve laid out the framework and categorized the different companies within the framework let’s have a look at a specific case to turn what we’ve learned into something tangible.
Content, before and after Facebook
Let’s have a look at the example we started with: the newspaper. Facebook has disrupted newspapers in a major way. Coming from the pre-internet world the editors were sure that the way their business worked was by creating a great product (the content) that people within their geography paid for because it was simply so good, and in addition the advertisers advertised in it because it was such an effective way for them to reach potential customers.
Then the internet came along and allowed anyone to distribute content. The supply of content, which was previously limited to how much the journalists in the newsroom could produce was now practically infinite. For a decade or so this was all quite messy and users would visit heaps of different blogs and sites to explore content online.
Facebook started out by bringing your offline relationships online. However, with the News Feed, Facebook made it easier to reach an audience within and outside your circles while on the other side giving everyone a personalized feed for discovering content. As opposed to newspapers, Facebook has the scale of content and number of users that enable them to invest huge amounts in technology infrastructure that enable these personalized experiences.
They can also deliver a superior ad product to advertisers which means their revenues can scale without marginal costs. All happening in a virtuous cycle that improves the experience for all parts (content creators, content viewers, and advertisers) as Facebook grows.
So it turns out that what the newspapers actually had more than anything was a monopoly on the distribution of content in their region. Readers and advertisers to a large extent paid publishers because they did not have alternatives. Now that Facebook owns the eye-balls (demand) and can serve an infinite amount of content most publishers are commoditized, lacking access to their potential customers and have an inferior ad offering for advertisers. A newspaper might not like this and stop posting on Facebook, but that void will simply be filled by an infinite amount of other content that would love access to Facebook user’s eyeballs. Simply put, the newspapers have been disrupted.
Aggregation Theory explains how these companies became so powerful and what sets them apart from pre-internet companies. Crucially, they are controlling the end-user demand and commoditizing their suppliers.
Note that these are all consumer-facing businesses. Businesses have more requirements for customization and thus won’t accept one-product-fits-all that the aggregators offer. In addition, they have more interest and bargaining power when it comes to protecting the usage of their data. These days venture capital is mostly deployed in B2B startups because it’s almost impossible to win against the Aggregators when it comes to reaching consumers. I’ll quote Ben directly here:
What is important to note is that in all of these examples there are strong winner-take-all effects. All of the examples I listed are not only capable of serving all consumers/users, but they also become better services the more consumers/users they serve — and they are all capable of serving every consumer/user on earth. This, above all else, is why consumer technology companies are so highly valued both in the public and private markets.
As the scrutiny and skepticism towards the Aggregators keep rising, Ben’s framework is a great tool for understanding how one might want to curb these companies. This is something Ben is writing a lot about these days, and hopefully, policymakers will learn and listen instead of simply applying pre-internet economic models to internet-enabled businesses.
I hope you found this post useful and inspires you to explore more of Ben’s great work. Feel free to ping me on Twitter @hagaetc with any feedback, questions or thoughts.
Have a great day!
Sources and further reading
Articles underpinning this post (list is probably non-exhaustive):