Copyright: bplanet

How do platforms compete? And with whom do they compete? The near-impenetrable wording of the new Digital Markets Act of the EU suggests that regulators want to regulate platforms, but do not have the conceptual model to phrase it clearly. As pointed out by Fiona Scott Morton and Cristina Caffara [i], each prohibition in the Act seems to refer to a past court case. The DMA lacks a unifying vision that explains why platforms must be regulated this way. This vision, they say, must be expressed in terms of platform business models.

In this blog I will identify the characteristic features of platform business models that explains why regulation is needed. These features are, first, that digital platforms not only interact with their customers, but also with their customer’s’ customers, and second, that digital service provision equals digital data collection.

Let me start by showing the general pattern.

Platforms are layers in a digital infrastructure

A platform is an infrastructure in a digital value network on top of which others can create additional value. For example, a digital marketplace is a platform on top of which third-party sellers can offer products to consumers; an operating system is a platform on top of which app developers can offer apps to end users.

In general, the value-adding parties of a platform are called complementors, and the users of the added value are called end users. The characteristic value-adding hierarchy of platforms shows that platforms provide services to end users as well as to complementors.[ii]

For example, the complementors of a marketplace are third-party sellers. The complementors of an operating system are developers. For search platforms, complementors are publishers of web pages, and end users are readers of web pages.

In social networks, complementors and end users are two roles of the same group of people, who alternate between publishing and reading posts.

Ad-funded platforms have a third side, advertisers, who want to attract the attention of end users. The platform matches content shown to end users with ads provided by advertisers. I will show the value hierarchy of ad-funded platforms below.

Differences with product families

Platforms and their complements together form a value hierarchy that is similar to product families.[iii] A product family like a line of cars has a common architecture, on top of which the manufacturer adds useful complements to create a car. The common architecture and the set of complements are designed by the manufacturer, and they are designed before the product family is manufactured.

The difference is that in a platform value hierarchy, the complements are added by others. They are not designed by a single entity, and they are not designed before the platform hierarchy is implemented. Instead,

• the platform invites commercial entities to add complements and
• the platform value hierarchy grows organically.

The platform may create some complements itself, but there are numerous competing entities that add complements too during the lifetime of the platform.

This creates different incentives to define platform architecture. In a product family, the manufacturer will include shared functionality in the common architecture but in a platform value hierarchy, the platform provider will include functions that generate most revenue in the platform architecture. The platform will imitate the functionality of successful complements in complements of its own. As a result, complementors not only compete with each other, but also with the platform.

Digital service provision equals digital data collection

There is a second feature that warrants special regulation foe digital platforms. This is that in the digital economy, to provide a service is to collect data about usage.

Complementors and end users provide the platform with data each time they use the platform. This gives the platform data advantages that create an unequal playing field. The following value hierarchies characterize this unequal playing field.

Digital marketplace competition patterns

Online marketplaces host retailers who sell products to buyers and provide useful search and transaction functionality to buyers. Here is the value-adding hierarchy.

This is similar to a shopping mall that hosts retailers and provides some services to buyers. However, a digital marketplace in addition performs transaction services to all retailers and buyers. It offers a single search function that searches the catalogs of all retailers and provides one payment function that must be used by all retailers.

This creates a powerful information position for the marketplace. Here is our first platform competition pattern.

Marketplace MFN. Since the marketplace knows the prices of all products of all retailers, and can monitor their prices elsewhere, it can demand that retailers do not offer their products cheaper elsewhere. The marketplace must be the “Most Favored Nation” (MFN) of each retailer.

Amazon Marketplace used to demand this from their sellers, but it dropped this requirement after regulatory scrutiny. However, a marketplace can achieve the same result by moving products sold cheaper elsewhere down on the list of product search results. Those products will then never be found by buyers.

Owning the product search function used by all buyers on the marketplace opens another source of revenue.

Sponsored search results. The marketplace function may accept payments from retailers to list their products highest in the results page. To be listed on the first results page, a retailer must pay more than those listed on the second and later pages.

In this way, a digital marketplace creates a competitive environment for retailers who must list their products cheaper than elsewhere and must pay the marketplace more than other retailers to get found quickly by buyers.

A third competition pattern in digital marketplaces is that the marketplace itself can own one of the retailers. Amazon created this configuration when it opened its retail infrastructure to third -party sellers in 2001. The marketplace then hosts two kinds of retailers, its own platform retailer and all others, now called third-party sellers.

Platform retailer. A digital marketplace may own one of the retailers it hosts.

In this pattern, the marketplace owner competes with third-party sellers also hosted by the marketplace. This would be similar to a shopping mall who owns a shop in the mall.

Because the marketplace performs all payment transactions of all retailers, the platform retailer has an information advantage that allows it to sell popular products at lower prices than the competition. This advantage is amplified by the third and fourth marketplace patterns.

Platform-labeled products. Using the information of the payment transactions on the marketplace, the platform can manufacture popular products under its own label and sell them through its retailer at a competing price.

Platform self-preference. The product search function may preference platform-labeled products in its search results list.

The DMA tries to prohibit some of these practices, in particular the Marketplace MFN, platform retailer, and platform self-preference patterns.

Operating system competition patterns

Operating systems are platforms that offer services to app developers and app users, as shown in the following diagram.

Just as a marketplace can host a retailer that it owns, an operating system can be a platform for a developer that it owns. The OS provider is then a developer of apps for its own OS.

This gives the provider a powerful position with respect to competing app developers. It enables the following competition patterns.

App defaults. The OS provider can pre-install its own apps, set them as defaults, and make them non-removable.

Interoperability barriers. The OS provider can make it impossible for competing apps to run on its OS. For example, it can ensure that only its own payment app runs on its OS.

Switching barriers. The OS provider can raise a barrier to switching to other apps, for example by decreasing the cost of purchases inside its own app.

The DMA tries to prohibit these three competition patterns.

App store competition patterns

An app store is a marketplace for apps, that itself runs as an app on the operating system. App stores combine the value hierarchies of marketplaces and operating systems.

App stores exhibit the competition patterns of marketplaces and operating systems, and some more. They are pre-installed, default and non-removable apps that do not allow competition. They can preference its own apps, i.e. the apps developed by the OS provider, in the app search function.

In addition, the OS provider has the following three app store competition patterns in its toolbox.

App exclusion. The OS provider can refuse to run apps not acquired through its app store.

Native payment preference. App developers can be forced to use the native payment app of the IOS provider in in-app purchases even after the app has been downloaded and installed. This would be a condition for availability in the app store. It would give the OS provider important data about app popularity and usage, which can be used to develop popular apps itself.

Transaction fee. The native payment app used for all transactions in an app could charge a high transaction fee, say 30% on each transaction.

The DMA forbids all three competition patterns.

Ad-funded platform competition patterns

The fourth group of competition patterns prohibited by the DMA is about ad-funded platform revenue models. For example, an ad-funded search platform may place ads next to search results, and an ad-funded social network platform may place ads next to posts. In both cases, the platform matches content with readers, and matches inventory in this content with advertisers. “Inventory” is advertising jargon for empty space on a page where an ad can be shown.

The following diagram shows the ad-funded value hierarchy. Publishers in this pattern are publishers of web pages or authors of posts.

The two rounded boxes at the bottom of this diagram are not business entities, but activities of business entities. Matching inventory to ads is an activity in which advertisers buy inventory (ad space) to fill it with ads targeted to an audience of readers. Examples are Google AdSense and Facebook Audience network, which aggregate inventory of several publishers, and Google Ad Manager, which conducts an auction in which advertisers bid on inventory.

Matching inventory to ads generates revenue that is used to pay for the other activity, matching content to readers, which is free for readers.

This creates a value hierarchy in which a platform may compete with its complementors. When a platform matches content to readers, it creates inventory in which it can show ads. When it matches inventory to ads, it can choose to sell its own inventory or that of the third-party publishers in its ecosystem.

This is the first competition pattern in the ad-funded platform revenue model.

Ad-funded platform publisher. An ad-funded platform that matches content to readers can sell own inventory as well as that of third-party publishers.

This is similar to the platform patterns that we saw earlier, namely a marketplace that runs a retailer, and an operating system provider that develops apps. However, the DMA does not mention the ad-funded version of this pattern.

Two other patterns in ad-funded business models have to do with promoting data collection by the platform and decreasing data provision to its customers.

Ad-funded platform reader identification. An ad-funded platform can force publishers and advertisers to use the platform’s reader identification system. This promotes data collection by the platform and hinders data collection by its customers, the publishers and advertisers.

Obscure pricing and performance. An ad-funded platform can create complicated pricing schemes and prevent independent auditing of ad performance.

The DMA contains several articles that prohibit these two patterns.

Lessons learned

Platform business models are layer cakes in which the platform provider tends to play a role at several layers. The competition patterns identified above are enabled by the fact that platforms invite complementors, who compete with each other and with the platform, and by the fact that digital platform services are data collection instruments.

The patterns also illustrate that the details of the business model may be hidden in implementations. How does product search work? What data does a marketplace collect about its transactions? What are the interoperability requirements of an app store? How is inventory aggregated? How are inventory auctions conducted?

To write clear, generalizable regulations for the platform economy, we need to understand platform business models. This will help us identify the technology decisions that facilitate unfair competition. In the next blog, I will zoom out and consider all four aspects of business models: value proposition, value hierarchy., revenue model and delivery model.

[i] Fiona Scott Morton & Cristina Caffarra. “The European Commission Digital markets Act: A Translation.” VoxEU/CEPR. 5^th January 2021. https://cepr.org/voxeu/columns/european-commission-digital-markets-act-translation

[ii]  Roel Wieringa & Jaap Gordijn. Digital Business Ecosystems. How to Create, Deliver and Capture Value in Business Networks. TVE Press, to appear, 2023.

[iii] C. Baldwin & C. Woodard. “The Architecture of Platforms: A Unified View.” In A. Gawer (ed.), Platforms, Markets and Innovation. Edward Elgar Publishing, 2009. Pages 19-44.