• The ability to match demand and supply (allowing for mutual discoverability) thanks to their ownership of demand is only one of the two advantages of aggregators, the other being the ability to allow counterparts who do not know each other to transact with each other.
  • In the markets where the supply of aggregators operate, centralization was caused by the demand using popularity of suppliers as a proxy for their trustworthiness.
  • As a consequence of the fact that aggregators can provide the otherwise missing trust element between a customer and a product or service provider who do not know each other, and thus popularity becomes a negligible factor in the users’ choice of a provider, the distribution of profits in markets in which the supply of aggregators operate flattens over time, whereas the market at the above layer (the aggregators’) centralizes.
  • The importance of the trust element opens up the possibility of a new threat to aggregators: whereas their self-reinforcing dominant position on the match-making side is hard to be disrupted, this isn’t true for their position on the trust-enablers one. In particular, the share of their profits that come from replacing costs to act in trust-less environments is open for disruption from any new entrant with an innovative trust-minimization technology, such as dry technologies.

Note that I do not consider the centralization that derives from current aggregators any good (because it creates other risks, unless such aggregators are decentralized; more on this later).

(The concept of trust in aggregation theory is not novel; Ben Thompson already wrote in 2015: “Airbnb, for example, deals with vacant rooms; what makes it work is the way it has digitized — and thus commoditized — trust. Uber deals with cars; it has digitized both trust and dispatch.”; in this essay, I’m expanding the concept with a focus on the effect of trust on the distribution of profits.)

Aggregation Theory

(If you are familiar with Aggregation Theory, you can skip this and jump to the next part of the essay, “Trust and Centralization”)

The digital age saw the rise of aggregators, such as Google, Facebook, AirBnB and Uber.

In his Stratechery, Ben Thompson defines aggregators as companies which possess three characteristics:

  • A direct relationship with users,
  • Zero marginal costs for serving users,
  • Demand-driven Multi-sided Networks with Decreasing Acquisition Costs

For example, in the case of AirBnB:

  • Users directly interact with AirBnB’s app or website.
  • It doesn’t cost anything to AirBnB to serve one more user, as it doesn’t own nor ship the apartments.
  • When AirBnB was small, it was quite expensive to get a new tenant to list his apartment on the website (the founders were going door-to-door!); as it grew, tenants were progressively more incentivized to join the website as that was where most users would go to look for an apartment to rent.

Trust and Centralization

There might be many reasons why many markets become centralized over time: compounding, economies of scale and network effects are the most common. However, these are not determinant in any of the markets where the suppliers of aggregators operate. Here are some examples:

  • Hospitality: an hotel has negligible economies of scale, as the costs of the land, the building and the furniture are the determinant factors. Similarly, an hotel has negligible network effects. However, the reason why a popular franchise hotel is very profitable is that, while traveling to a city they do not know, many customers book a room there because they know what to expect; in other words, they trust the hotel to deliver on their promises.
  • Ride-hailing: ten years ago, taxis where the only not outrageously expensive choice for people who needed a ride somewhere and did not own a car or could not take public transportation. However, calling a taxi is not less expensive than paid hitch-hiking, nor, in some geographies, it is more convenient. (In Kazakhstan, for example, it is faster and easier to just stand on the side of the street, raise your arm and pay a small fee to any of the 2-3 cars which will stop within one minute. This is much faster than summoning a car using an Uber-like app such as Yandex, which usually takes a few minutes.) The main reason people used to call a taxi is because they could trust the driver.

(Stratechery readers might have noticed that all examples so far only include level-2 aggregators. What about level-1 and level-3 ones? I will talk about them in the appendix, not to overcomplicate the core essay.)

If you are not convinced yet of the importance of trust in the aggregators’ markets, imagine traveling back in time 10 years and telling your past self that in 2008 you can easily find from your laptop a room in someone’s house for $20 (AirBnB) or use your phone to get someone to ride you to the airport for half the price of the taxi (Uber). Your past self will not so be so surprised from the fact that such services are available on your phone or that they cost so little, but mostly by the fact that you are sleeping at someone’s you don’t know and that you are riding in a stranger’s car.

In the absence of the possibility to quickly assess the trustworthiness of a supplier they do not know, customers use popularity as a proxy. In such contexts, everyone wants to buy from the same few suppliers whose name they heard, which in turn become even more popular. As a results, in markets without aggregators, profits concentrate in the hands of the few popular suppliers.

Aggregators and Trust

Aggregators allow counterparts who do not know each other, and would therefore not trust each other, to engage in business together without having to pay extra costs such as vetting each other, signing contracts, purchasing insurances or involving lawyers. AirBnB, for example, allows both the guests to trust the hosts and the hosts to trust the guest (thanks to a 2-ways review system, a theft insurance, a payment system where the guests pay in advance, and a credit/reimbursement system).

In an environment where a customer does not need to know his service provider in order to trust him, there is no more need to resort to popularity as a proxy for trustworthiness; therefore, popular suppliers do not capture anymore a disproportionate share of the profits.

Therefore, aggregators tend to flatten the profit distribution of the markets where their suppliers operate, while at the same time capturing most of the profits of the markets where the aggregators themselves operate. Hotels are increasingly less the only choice for accommodation while traveling, but aggregators such as AirBnB and <> are increasingly more the best choice for finding an accommodation.

The presence of an aggregator transfers the trustworthiness burden, and the profits, to the upper layer. No more do the suppliers have to prove their trust; the aggregator has to be trusted by the demand; no more do the supplier collect a disproportionate size of the profits in relation to their capacity to be popular; the aggregator does.

Of course, the aggregators’ ability to make its suppliers behave without breaking trust is that there is too strong of a penalty for a supplier to break the trust: delisting. For example, an apartment owner who lists his rooms on AirBnB has very strong incentives not to break trust, for otherwise he might get bad reviews and be effectively out of the only relevant marketplace for short-term shared renting. However, it is not enough that the aggregator can make sure that its supply will not break trust; the aggregator has also to be trusted by its demand.

The Future of Aggregators

The profitability of aggregators is based upon two points: to be popular amongst the users, and to be trusted to the point of minimizing the trust requirements for transactions to happen. (The importance of the former point dwarves the importance of the latter, though the latter can allow to capture profits from time and money which would otherwise have to be allocated to other activities, such as insurances, vetting, contracts and lawyers).

The first point, popularity amongst users, is self-reinforcing and very hard to attack by new entrants. However, the second one is ripe for innovation. Any new entrant who could devise of a way to allowing for even safer transactions between parties who do not know each other might be able to take an aggregator’s place (unlikely, due to their ownership of demand 1) or at least might take the share of its profits that comes from trust-minimization, while leaving to it the share that comes from match-making). 2.

Let’s imagine, for example, that a smart door-lock is invented which allows a landlord to rend half of his house to a guest, and to ensure the guest that the landlord cannot access that half of the house. Three scenarios open:

  1. A new platform develops around houses having installed the smart door-lock, and travelers have the choice to either use AirBnB to rent “unsafe” rooms or the new platform to rent “safe rooms”. This scenario is unlikely, because the new platform would have a demand problem – which users would use the new platform if AirBnB has orders of magnitude more listings? And which landlord would not want to list his rooms for rent on AirBnB, where the demand is?
  2. Landlords who installed the smart door-lock list their houses on AirBnB and collect a premium, because they can charge more for a safer solution (most likely scenario).
  3. Landlords who installed the smart door-lock list their houses on AirBnB. Meanwhile, the smart door-lock manufacturer, or a third party, create a website which lists the positions of all houses with a smart door-lock and offers them for rent, and makes the website so convenient to use that it isn’t a big cost for users to look for listings on both AirBnb and their website.

To be clear: I would say that scenario number 2 is the most probable. However, if someone told me that in 20 years someone managed to challenge the position of an existing aggregator, I would say that the most likely cause is scenario number 3: a provider of a trust-minimization technology managed to challenge the popularity of the aggregator, recognizing that popularity is mostly a proxy for trust. In particular, I’m curious to see how smart contracts will influence the current aggregators.


Level-3 aggregators

Stratechery readers will have noticed that this essay mostly talked about level-2 aggregators (those who have some supply transaction costs, such as Uber who has to vet its drivers), and did not mention level-3 aggregators so far (those with no transaction costs at all, such as Google or Facebook).

I would argue that there is a very important trust element for level-3 aggregators as well. An advertiser who advertises on Google and Facebook trusts, somehow, that the impressions their ad receives will be towards users who are potentially interested into the ad, thanks to Google and Facebook’s superior targeting capabilities.

Aggregators, Trust & Monopolies

Are aggregators monopolies? Should they be permitted to censor? My not-fully-formed opinion is: aggregators are monopolies and they should not be permitted to censor.

Those who argue that aggregators are not monopolies usually bring the argument that monopolies usually have a monopoly on shelf space; aggregators are digital, and in the digital world the shelf space (on the internet) is unlimited. Sure, the digital shelf space is unlimited, but the trusted shelves are limited. In markets where trust is relevant,

You can find more essays of mine at <>

Also published on Medium.

  1. I’m curious, though, about what would happen in case a series of shocking events would hit an aggregator, such as some AirBnB tenants killing their guests; would the sentiment shift so that a less popular though more trustworthy solution could be preferred by at least part of the users?
  2. In particular, dry technologies (techs who do not require human action) such as smart contracts or machine-operated door locks and self-driven cars could be potential enablers of such trust-minimization

By admin

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.