Monday, 20 May 2019

The problem with Ben Thompson's 'aggregation theory'

Ben Thompson, via his website Stratechery ('Strategy' and 'Tech'), has made a name for himself over the past decade popularising the concept of 'aggregation theory'. I invite readers to check out his material directly for a fuller explanation, but in short, his argument is that tech platforms have become so powerful (and profitable or potentially profitable) because in the internet era, the game has changed, and it has become far more important for companies to aggregate (and control) demand than to control supply (the latter of which Ben argues was the case in the past). Organisations that now control demand are the ones destined to prosper, and those companies are the large tech platforms.

Examples of demand aggregators are Netflix (that aggregates viewers of scripted content), Amazon (that aggregates buyers of (e)books and general online merchandise), Google (that aggregates internet search, and operates a toll booth redirecting traffic to the highest bidders), Uber (that aggregate people looking for taxi services) and travel and airline ticket booking companies such as Expedia,, Webjet, and AirBnB (that aggregate people looking to book flights and hotels/accommodation). The argument is, people that control demand are now more powerful than those that control supply, because if you control demand, you dictate what supply is able to receive any demand (i.e. what supply has market access), and therefore you make all the money.

Superficially, this theory seems to have a lot of explanatory power. Take the example of hotel booking platforms. These organisations are able to centralize the demand of people looking to book accommodation, and direct that traffic to available sources of supply. If you own a hotel and aren't on the platform, you can't reach buyers any more, and so your only choice is to use the platform and pay Expedia (for e.g.) a huge cut ('take rate') to make bookings on your behalf.

The problem I have with the theory is that it implies there is something fundamentally new or unique about the economics of the brave-new-world of tech, when in reality, the old economic rules still work just fine. This, in turn, creates the raw material to rationalize bubble thinking/valuations, instead of more level-headed analysis. The reality is that from time immemorial, it has always been the case that certain points in the supply chain make more money than others, reflecting differences in market power. Porter's Five Forces, for instance, has long been used as a framework for analysing where and how much market power exists, and explaining and predicting why some firms make more money than others. If your suppliers for e.g. have a lot of bargaining power, all else held constant, you tend to be less profitable, and vice-versa.

There isn't a lot that is novel about that insight, and the truth is, contrary to what Ben argues, demand and supply have always been of equal importance, and remain so. The real reason certain platform companies have become powerful is not because demand has become more important than supply, but because in many cases, the demand side has become more concentrated than the supply side, thereby giving it more market power. Demand has therefore become more 'consolidated' (an old-hat term which I prefer to the new-hat 'aggregated') on the demand side than supply has on the supply side, but this is not the first time that has ever happened. Carnegie and Rockefeller got rich by consolidating their industry and controlling the infrastructure that allowed both suppliers and competitors to access the market. If you 'aggregate' for e.g. all the railways that provide access to market, surprise surprise - you will make a lot of money (in this respect, 'aggregation' is just a synonym for monopoly). I also recently wrote a blog post about how PBMs (Pharmacy Benefit Managers), working in conjunction with health insurers, were getting rich by controlling which pharmaceutical products were able to achieve reimbursement (and hence have market access).

If we consider the example of travel booking platforms like Expedia and, after years of mergers/consolidation (and inadequate regulatory antitrust oversight, as the regulators have not understood the importance of regulating 'platform' market power), there are now fewer mainstream travel booking sites than there are hotel chains or airlines. Consequently, it ought to be little surprise that booking websites currently wield greater market power. People thought it was going that way with Netflix as well - you had one organisation apparently destined to control demand/eyeballs, whereas there was a larger and more fragmented number of organisations supplying content.

However, not all tech sectors are like that, and nor is it inevitable that the supply-side will always remain much weaker than the demand-side in many other segments. An example of an industry that is different is music streaming. In the world of music, supply is still more aggregated than demand. Three major studios control virtually all commercial music, whereas at present there are more than three major sources of platform demand (Spotify, Pandora, Apple Music, YouTube, and Amazon), as well as conventional ad-supported radio. Because supply is more aggregated than demand, music streamers aren't yet making any money, but the big music labels remain very profitable. In Ben Thompson's parlance, it could be said that the reason is 'supply aggregation theory'. In conventional parlance, it is because the supply-side is the point in the supply chain with greater market power.

Furthermore, in competitive free market economies, there is always an evolving competitive jostle for a greater share of the spoils, and 'aggregation theory' implies a degree of structural permanence that I do not believe should be automatically assumed. That, in turn, creates the risk of a lack of vigilance about some of the risks of change. When one part of the supply chain is profiting at the other part's expense, given enough time, one or both of the following things can happen: (1) the portion of the supply chain that is being under-remunerated has an incentive try to adapt and improve its bargaining position, including via market consolidation; and (2) disruptive competitive pressures are also usually eventually brought to bear on points of the supply chain that are over-earning relative to the level of their value-add, and 'new' platform businesses will be no more immune to this in the long term than 'old' businesses have been (after all, in the long term, the new becomes the old). And at present, many platforms are currently materially over-earning.

I listened to a podcast not so long ago where one of the tech-infatuated guests was arguing that tech is taking over the world and increasingly supplying everything we need, noting that on a recent visit to Sydney, he booked his flight through Skyscanner (Expedia); took an Uber for transport from the airport; stayed in AirBnB accommodation; and booked food delivery through UberEats (if memory serves). He also found activities to do through TripAdvisor, while booking restaurants through OpenTable. The implication was, how we live, work, and play is now being completely overrun and serviced by tech, and so of course tech stocks are increasingly conquering the world.

However, it doesn't take much reflection to realise that it is actually the real world, not the world of technology, that is providing all these services. All those apps are doing is providing an algorithm that lowers search costs and makes booking easy. Expedia didn't design, build and maintain the airplane that flew him to Sydney; build or operate the airport; train pilots; or find, produce, refine and transport the necessary jet fuel to power the plane over its continental voyage. Uber didn't design and manufacture the car used to transport him to his hotel; find, produce, and process the raw materials that go into it (such as steel and aluminium); or actually drive him from the airport to his hotel. AirBnB didn't design, build, maintain, or clean the house he stayed in, nor supply it with electricity. UberEats and OpenTable didn't grow and process any raw foodstuffs, or use them to cook a meal, and TripAdvisor didn't design, manufacture or operate any of the tourist attractions he visited.

In fact, all these companies did was write some pretty simple code that made matching buyers with sellers easier and more efficient, and the real question that should be being asked is whether these platform companies are extracting too much value from the supply chain relative to their value-add, and whether that is likely to be a sustainable situation in the long term, or will invite potential disruption and/or an eventual supply-side/regulatory response.

Take Webjet, for instance - and Australian flight booking company. The company takes about $30-50 per flight booking, which can easily be 10% or more of the price of the flight. Ask yourself, is Webjet really providing 10% of the value add? What about the pilots and flight staff; the maintenance engineers; the petroleum engineers finding, producing, and refining fuel; and the manufacturers of the aeroplanes, as well as their financiers (leasing cos)? The truth is, Webjet is actually providing something more akin to perhaps 1% of the realistic value add, and so it is therefore currently significantly overearning. It's like a stock-broker charging a 10% commission on a trade. Stock brokers also used to charge enormous commissions, but they were significantly overearning, and the forces of competition therefore lowered it over time, to levels that today are practically zero.

In capitalism, what is fair is irrelevant, however. Only market power matters. It doesn't matter how much money Webjet 'should' be making - only how much it is actually able to make. Because the airline industry is more fragmented than the travel booking market, Webjet currently has more power - it has indeed 'aggregated' more demand than airlines have managed to aggregate supply.

The question that should be being asked at the moment, however, is what happens next? Is the current situation sustainable long term? Or is it inviting disruption, with booking rates destined to be competed down to levels closer to their real value-add in the long term? There are many ways in which this might happen. There will of course always be new entrants attempting to enter Webjet's point in the supply chain, and provide a better service at a lower cost with increased convenience. For instance, perhaps a new start-up comes along that can use AI to automatically book you a flight using voice-activated instructions (maybe an Alexa app) - indeed maybe Google or Amazon comes up with such a service. If it's faster and cheaper, Webjet will quickly be displaced, and it's fat margins create a significant opportunity of anyone able to find a way to successfully supplant them.

However, another potential mechanism is for the supply-side to respond and improve its bargaining position, so it can recapture the value currently being ceded to demand-side aggregators. This supply-side response has already begun in certain industries. In the traditional media industry, for instance, studios have begun to aggregate supply to fight back against Netflix. Several megamergers have already happened, which have reduced the number of organisations controlling scripted content production, and content libraries. If they can aggregate supply in the manner the music industry has, while also supporting alternative streaming platforms, the power of Netflix will be undermined.

Other industries should learn from this. Major hotel groups should merge and/or co-operate and form their own hotel booking site, and either withdraw their supply from booking platforms, or (more realistically) continue to make them available, but at higher prices than on their booking sites (easy do do by cutting out aggregators' take-rates). Where applicable, they should also lobby to have archaic restrictions on the ability of hotels to offer prices on their own website lower than on external platforms. They will then be able to offer lower direct-booking prices to customers, allowing customers to find a room they like on an external platform, and then book the room directly and more cheaply on their website. 

In addition, new tech organisations are likely to spring up that provide booking tools that allow customers to more easily contract with independent operators without having to re-insert their credit card and identification details each and every time. If someone could create an ID authentication/payment plug-in app that enabled one-click purchasing on a variety of different merchants' sites, it would help independents' websites disintermediate the aggregators (this is a major business opportunity for somebody out there - I'm not going to do it so feel free to steal my idea).  

Airlines should also join forces and create their own flight booking platforms, put all their collective inventory on it, and then withdraw the supply from external aggregators (or charge Webjet et al significant price premiums to carry the same inventory). They should then advertise their lower rates to customers. In reality, the suppliers ought to have much more power than the aggregators, because Webjet doesn't have any of its own planes (unlike Netflix which can self-produce its own content). That the aggregators are making more money than the airlines is stupid. The dominant booking platform should be owned by the airlines. If they were smart enough and got their act together, they could capture this value themselves - use their 'aggregation of supply' to squeeze out demand-side rent-seekers that are earning huge amounts of money but adding very little value.

Different industries will have differing degrees of success with this, and it won't happen quickly, but tech investors should not be complacent. Tech platforms have been disruptive, but the history of capitalism shows that it is not long before the distruptors become the disrupted. In the very long term, replacement cost is the best estimate of value, and many of these tech companies sport huge valuations despite having technology that is very easily replicated. In some cases network effects exist, that mean the replacement cost is in fact very high (it would in fact cost hundreds of billions of dollars to displace Facebook's 2bn users by paying them to migrate platforms, highlighting the power of the network effect), but only a small minority of businesses fall into this category. What's the replacement cost of a flight booking app? It's ultimately just a bunch of code. 

In short, aggregation theory doesn't really add a whole lot to the analysis, in my view, other than simply restating the uncontroversial view that the balance between demand and supply matters, and that the players in the supply chain with the most market power will make the most money. These are not novel concepts, and there are too many 'this time is different' and 'the old economic rules don't apply to tech' connotations to it for my tastes. This sort of thinking is what lead to the 1998-00 bubble. It burst and people realised, actually the old rules still do apply after all.

It is going to be interesting to see how things evolve in coming years, but I would counsel caution about paying up huge multiples for platform/booking businesses with seemingly assured future rapid growth and prosperity. History shows that the future tends to surprise people, and that today's sure winners often prove far more susceptible to market forces/competition than is generally believed during boom periods.



  1. This comment has been removed by the author.

  2. Very interesting. I wonder, though, if antitrust law would permit airlines, hotel companies, etc. to join together to operate a unified booking site for consumers.

    1. Agree, but on the flip side (something i forgot to mention in the article), there is also a ‘risk’ regulators eventually take action to reduce the market power of/regulate de facto monopoly platforms as well. As was the case with Carnegie, Rockefeller, and indeed Microsoft in the late 1990s, regulators eventually responded to their monopoly control of demand with regulatory actions/break ups that reduced their power.

  3. Hector Mustache20 May 2019 at 17:29

    Interesting insights Mr Taylor. I agree. The tech kids should read Competition demystified from Greenwald. Right now I like most Teekay LNG, Fairfax India, Raiffeisen Bank and ITE Group. I would love to find one undervalued company in China. There must be plenty.

  4. Great post. I remember in the mid 2000's, I think it was, when the zeitgeist was "dis-intermediation". The internet meant suppliers would reach out to consumers directly, cut out the middle man, and with all power shifting to them, they would march off to a low-cost sunset, hand-in-hand with happy customers. I remember much banter about this on Bristlemouth, the Intelligent Investor's associated blog site, back then.

    Fast forward to today, and we have intermediary behemoths that we call "tech companies", and the new zeitgeist is that these will forever hold all the power.

    Funny how we go from pillar to post when we allow ourselves to believe the consensus.

    On the subject of the risks WEB faces of being dis-intermediated, there is some irony that an old world intermediary like FLT looks at less risk of such (in my opinion). FLT is a distribution channel that carries high fixed costs. There is no rational reason, that I can see, that an industry with pronounced demand/inventory cycles, would want to permanently carry that cost & complexity. FLT offers a channel for airlines to off load excess inventory when/if required. In short, from the suppliers point of view, FLT has a reason to exist. In agreement with you, I'm not sure the same could be said about WEB.

    1. Great insights. Is indeed interesting how the narrative on disintermediation has shifted. As you note, we now simply have new intermediaries in place of the old.

      The more successful ones are really to a large extent simply 'marketing' organisations that get paid to create superior visibility for one supplier ahead of others, in a crowded field of suppliers. In that respect they are rentier business models - they are intermediaries that have inserted themselves between buyers and sellers, and are capturing a lot of the value provided by suppliers.

      Hard to see right now what the mechanism is - because it always is - but rentier middle-man business models usually get disrupted at some point, and the continuing evolution of technology, including AWS etc that keeps lowing the barrier to entry for new tech start up, suggest they probably will be at some point. If nothing else, there is certainly a clear incentive for suppliers to move to another platform that shares the spoils with them more equitably.

  5. The examples given of companies that control demand were about to do so because of significant disruptive forces (the advent of railways; the internet). Absent regulatory intervention or supply side competitors working together, it may well take another significant disruptive force for these companies to be displaced. The chance of a significant disruptive force emerging is, surely, small. And the chance that that force is in the relevant industry must be even smaller. I would suggest these companies are safer than the impression given in this post. Great thought provoking post, though.

  6. I recently stumbled across your excellent blog. Are you based in Singapore and still managing some outside money? Do you have an email contact?

    1. Yes indeed I am on both counts. Feel free to email me at

  7. Surely your comment here has already been done by Paypal, right? :-)
    'If someone could create an ID authentication/payment plug-in app that enabled one-click purchasing on a variety of different merchants' sites, it would help independents' websites disintermediate the aggregators (this is a major business opportunity for somebody out there - I'm not going to do it so feel free to steal my idea).'

  8. Just to add a little historical insight - the airlines (at least the US airlines) already did "join forces and create their own flight booking platforms" 20 years ago. This was Orbitz, and it had exactly the impact you would expect - greatly reducing distribution costs in the industry via increased competition. The reality is that travel booking sites don't make any of their money from airlines, they make it all from hotels (though I can't speak to Webjet / the Australian market), and it's for this reason.

    The major hotels have actually already done the same (it's called though with limited success for a number of reasons - including the amount of fragmentation in the industry, which goes to the overall point. They also already make their supply "available, but at higher prices than on their booking sites" - see the big direct booking campaigns by Marriott and Hilton - but at the end of the day they have limited power to do so because even these biggest of brands account for <5% of the market for accommodations.

  9. Intersting critic, however I think the bit you may have missed in understanding aggregation theory is that the internet has made distribution free - a tectonic shift in the world. Now you can scale customers and collect fees across borders without needing physical customer infrastructure like retail branches.

    For an example, I recently worked in a large global bank that was being disrupted by an internet giant utilising this free distribution to reach customers and leveraging network effects to keep them hooked using masses of data for product and UX personalisation. Why couldn't we just build it ourselves? We tried and had some success in a single market. However, because we were shaped (org and culture) like a bank not a tech company it was impossible to respond to.

    This stuff, although looks like a simple website/app, isn't easy at scale and u will fail quickly without the right people and guess where they are all working? For large tech startups being paid the same salaries in some cases plus promises of multiples of upside traditional companies with their steady stock valuations simply can't compete with. We tried for 12 months to hire a CTO who could handle the challenge and got nowhere.

  10. I think you make some good points, but ultimately are being a bit too dismissive of the value-add of the demand side. Specifically, I think your comment "What's the replacement cost of a flight booking app? It's ultimately just a bunch of code" misses the mark in that the value of a flight booking app (or Airbnb or Expedia) is not the code, but the trust and attention that the app has earned from consumers. Code is necessary, but not sufficient, to succeed in today's "Attention Economy".

  11. Lovely blog. Thanks for sharing with us.This is so useful.

  12. Hi Lyall,
    Just want to point out that Skyscanner is not owned by Expedia, but by CTrip which in turn is partially owned by Booking Holdings.

  13. Hi Lyall, I am a fan of Ben Thompson. He replied to your post. I have always had a problem with one part of Aggregation Theory and I wonder what your thinking is on it.

    Ben says that the Internet enables zero distribution costs and zero transaction costs but is that real true? If so then why does Google and Facebook not have 100% gross margins? Why do these Aggregators have cost of goods sold if distribution and transaction costs are realy zero?

    Thank you for your article. I enjoyed it and I have added you and will read more of your work,

    ben nartman

    1. How different is that from when TV or radio was conceived? New viewers/listeners could tune in, across geographies, at minimal or zero marginal cost.

      Extremistan has existed for a long time.

  14. Buffet in 1998 in front of dot com entrepreneurs at sun valley:

    “It’s wonderful to promote new industries, because they are very
    promotable. It’s very hard to promote investment in a mundane product. It’s much easier to promote an esoteric product, even particularly one with losses, because there’s no quantitative guideline.”

    This was goring the audience directly, where it hurt. “But people will keep coming back to invest, you know.

    It reminds me a little of that story of the oil prospector who died and went to heaven. And St. Peter said, ‘Well, I checked you out, and you meet all of the qualifications. But there’s one problem.’ He said, ‘We have some tough zoning laws up here, and we keep all of the oil prospectors over in that pen. And as you
    can see, it is absolutely chock-full. There is no room for you.’

    “And the prospector said, ‘Do you mind if I just say four words?’

    “St. Peter said, ‘No harm in that.’

    “So the prospector cupped his hands and yells out, ‘Oil discovered in hell!’

    “And of course, the lock comes off the cage and all of the oil prospectors start heading right straight down.

    “St. Peter said, ‘That’s a pretty slick trick. So,’ he says, ‘go on in, make yourself at home. All the room in the world.’

    “The prospector paused for a minute, then said, ‘No, I think I’ll go along with the rest of the boys. There might be some truth to that rumor after all.’

    “Well, that’s the way people feel with stocks. It’s very easy to believe that there’s some truth to that rumor after all.”

    1. Thanks Arun - fantastic parable. Have heard it before but it's been a while so thanks for the refresher!

      I think it is also fair to say that it is more difficult to promote a more mature businesses with a long listed history of actual earnings/revenues over various cycles, because all these real earnings also constitute evidence of a company's limitations.

      In life, youthful ambition can be limitless, but subsequent outcomes necessarily need to anchor those ambitions back to reality, and the ratio of potential to outcomes must decline as people age. Likewise, early in a company's history, people are free to project all kinds of rosy scenarios/ambitions onto a company's future, but it becomes harder to do so as lofty ambitions need to be benchmarked against actual realities. This is often why you will find companies with long histories of strong and reliable profitablity priced below hyped growth stocks with little or no earnings. The former have demonstrated their practical limitations. The latter have not yet. When they do, however, more often than not, the reality falls short of people's hopes and dreams.


  15. (Reposting as this comment seems to have never appeared)

    Hi Lyall,

    I appreciate this note. I went ahead and wrote a response on Stratechery, and made it free for purposes of this comment:

    Naturally, I disagree with your conclusion, but not as much as you think! Appreciate the dialog.

    Ben Thompson

    1. Hi Ben,

      Thanks for your response, and I appreciate you taking the time to constructively engage. I also enjoy these types of debates, and I probably should have acknowledged in my original post that I've enjoyed reading/listening to your material in the past, and it has added value to my thought processes on certain issues.

      I agree that we share a lot of common ground, and that to some extent, the debate is semantic in nature. For instance, I agree with you that the aggregation of demand is what explains the power & profitability of certain platform companies. I'll also grant your point that I've only read a subset of your writings, and indeed did miss a couple of the posts you referenced (on Porters forces and Spotify).

      I guess where I'm coming from is that I've seen your aggregation theory increasingly used, without nuance, to justify bubble-equse valuations for a number of tech platforms, on the premise that any platform that has aggregated a large audience must necessarily be worth a fortune, to the point where the lack of any differentiated offering; risks of the evolution of industry market power; and/or the presence of huge operating losses, are being ignored. It's more than fair to say that you can't be personally blamed for people misusing the theory, however. I'd be interested in your thoughts on my recent Uber & Netflix posts, which are examples of the more practical application of my critique. I don't see these sorts of dynamic analysis done very often, because it is merely assumed ipso facto that because they are platforms with large audiences, their future prosperity/dominance is assured.

      So my disagreement is not really absolute, but a matter of degree. This is where the relevance of my point on the novelty of controlling demand gateways arises from. I think your example of Wal-Mart aggregating demand in small towns is an excellent one - better than the examples I used. Being the primary demand access choke point is not new.

      Now I agree that the internet has allowed this to be done with both a speed and scale that is unprecedented, but counterbalancing that is the relative ease with which those platforms can be replicated/displaced (with some important exceptions). The frictions of time and space in the real world have represented very durable sources of competitive advantage for companies able to corral demand. It remains to be seen whether online demand-corralling business models provide as durable and profitable. I suspect that in more cases than is currently believed, they will not.

      Best regards,

    2. Lyall, I just read your article on Spotify. It appears to me that you made the same argument as Ben's Aggregation Theory and even the crazy valuations based on it that you complained about herein this article.

      It seems you complained when you agree with Ben and even made the crazy valuation argument yourself.

      Yet you complained as follows:

      "This, in turn, creates the raw material to rationalize bubble thinking/valuations, instead of more level-headed analysis."

      But then you made the same valuation argument here:

      Lyall writes:

      On conventional valuation metrics, this stock looks very expensive, at 5x forward revenue, with the enterprise making consistent and sizable losses. However, alternative metrics suggest that there is still a plausible value-based bull case for owning the stock.

      so which is it?

    3. Anonymous,

      I can understand how it might at first seem inconsistent, but the point of my Spotify article was not that Spotify was a buy (I didn't buy it and still haven't), but rather in highlighting the limitations of traditional metrics in valuing certain scalable tech platforms. Spotify was simply a random company I used to highlight the point, merely because I happened to have been looking at that company earlier in the day. I could just as easily have picked several other companies.

      This post is about how one analyses such platforms and the durability of their potential long term earnings power. I don't deny that there are some great technology businesses with flywheel effects, where they have amassed considerable market power. My point in this article is to highlight that (1) just because you have aggregated eyeballs, doesn't automatically mean you'll make a lot of money; and perhaps most importantly (2) there is always a risk of unfavourable change.

      I do not believe the points made in these two articles are mutually incompatible with one another.


  16. Great article. Stopped paying for Stratechery long time ago after feeling it was mostly re-hashing common business sense in overly lengthy articles and nice drawings.

  17. I also struggled with the explanatory power of aggregation theory and the amount it's been referenced as a justification for the size of tech giants. I think the way you frame aggregation theory through the lense of market power is a much more accurate characterization of its usefulness. It's helpful but not a breakthrough idea.

    I do think you need to give it a little more credit though. I've come to think of aggregation theory more as defining the phenomenon where a company can accrue a tremendous amount of leverage over suppliers that are fragmented. The speed and scope of this leverage is enabled by the new capability for a company to add an additional suppliers at zero marfinal cost. It's the same game like you said but a new very effective strategy.

    Supplier consolidation decreases this market power but demand aggregators can still be very profitable, even if they're effectively a marketing funnel, especially as they add new services and features to their platform that make their products better. Airbnb experiences and Spotify's podcast integrations are good examples of this. It'll be exciting to see the next phase of these markets evolve.

  18. I’m going to read this. I’ll be sure to come back. thanks for sharing. and also This article gives the light in which we can observe the reality. this is very nice one and gives indepth information. thanks for this nice article... Frank

  19. My issue with BT's Aggregation Theory is that it is premised on demand now being aggregated because of zero marginal costs. However, the marginal cost to serve an incremental customer for Netflix, Google, Facebook etc is not zero. It is low to be sure but not zero. There are server costs, bandwidth, power, cooling, staff etc etc.