Netflix recently released the social dilemma – if you haven’t seen it already, change that fact ASAP – a documentary which examines the collection and utilisation of data on social media platforms and the impact this has on users. The documentary highlights the use of algorithms imbedded on these platforms, steering our actions through the use of targeted adds, specific media content and notifications. We could easily extend this to review AI used in ecommerce and eservices as well as social media. Together, these fall under the umbrella of double- and multi-sided platforms. the social dilemma is a well-presented deep-dive into the psychological drivers which have fuelled the development of technology to harvest and use data, for a high-level overview it’s accessible and interesting – seriously, check it out! The documentary does leave scope for questions though, one of which is “What (if anything) are regulators doing in response to this kind of technology?” The below aims to unpack what double- and multi-sided platforms are, how they work, the problems they pose from a legal perspective and how law is beginning to address these issues.

Double- and multi-sided platforms

Double- and multi-sided platforms are digital sites which offer platforms to different users. For example, Amazon is a platform for vendors and buyers; Facebook is a platform for advertisers and social media users; Booking.com is a platform for hospitality providers and holiday makers; AirBnB… you get the gist. Add a few big names like Google, Pintrest and Uber to the list and you begin to see the range of sectors and services these platforms cover.

One specialist interviewed on the social dilemma described the internet almost like a mall, which I find quite a helpful illustration as a starting point. When you enter Amazon’s online marketplace, imagine you’re entering a supermarket – you’re subject to CCTV and know that the store can watch your every move, there are advertisements and sales, there may be discounts and/or prospective rewards if you have a certain points-card. In large supermarkets, you’re likely to buy products all under one roof for ease rather than trawl across the mall to a competitor supermarket.

When you enter a double- or multi-sided platform, you hit that little “agree to all terms” button unthinkingly. You accept all cookies, having no idea what that means but suspecting that the only people who might offer you a free cookie are likely to be the staff at your local Pret if you look so exhausted and downtrodden that they think a cookie might cheer you up. You’re into the platform. However, the difference between any kind of brick-and-mortar supermarket and a digital platform is significant. A platform knows which products I buy, which products I add to my basket but subsequently remove, replace or request more of. A platform knows where my mouse hovers, for how long it hovers, which products I browse, for how long I look at those products, at what time of day I look at those products… Platforms can then knit together the data they collect on me as a whole. If in an evening I browse coffee machines, books on beginner Italian language-learning, holidays in central Europe, various companies which will deliver fresh pasta to my door, and bread recipes (all fairly typical go-to procrastinations for me, wild I know) a digital platform can gauge a reasonable view of me and begin to make assumptions on my preferences, disposable income and lifestyle. I soon find myself plagued by adverts for kitchen appliances, food delivery options or other foreign language books which I invariably indulge in browsing, fuelling the platform in two ways: (i) platforms achieve their aim when they keep me online because they are either aiming to present advertisements to me or they are aiming to sell me products, either way they aim to shape my behaviour and the chance of achieving that successfully increases the longer I keep scrolling; and (ii) with each click or hover, I offer up more information about myself to a platform which quickly learns which adverts, notifications and recommendations I am responsive to.

Impacts of data on business models

1. Information dissemination

The social dilemma looks exclusively at social media platforms. For social media platforms, data is used to improve and train algorithms which lengthen the time we spend on sites to share advertisements and content which shapes our behaviour and grows the social media platform. The impacts of this go far beyond my mild annoyance at being bombarded by adverts of coffee machines. The reality is that political agendas and fake news are widely distributed on social media platforms – for a UK example, have a look at Brexit: The uncivil war featuring Benedict Cumberbatch, a documentary on how targeted ads were used to gauge and shape the opinions of ordinary people and shape the leave campaign’s message.

If we look at double-and multi-sided platforms more generally, we can see that data is not just used to fuel better AI and disseminate information, it’s used to design tangible products and to quash competition in markets.

2. Killer acquisitions

“Killer acquisitions” are acquisitions by large platforms of relatively small targets which are insufficiently significant to trigger investigation by competition authorities but which play a significant role in extending platforms’ reach and dominance in specific markets. Competition authorities will traditionally step in if an acquisition is above a certain value threshold or if the two entities together would have a significant share of a particular market, though admittedly competition authorities are acting with increasing scrutiny. Killer acquisitions have typically fallen short of these thresholds.

The impact of these acquisitions has been to build platforms’ presence without triggering any material investigations. This means that in some markets, platforms gain huge competitive advantage, farming data and selling products, because of an increased platform from which to analyse consumers, and defeating potential competitors in the market. This has been a key focus, for example, in the pharmaceuticals sector where buyers may acquire smaller drug developers to benefit from their innovations rather than face these firms as a competitor.

3. Matching prices and products

A platform has the ability to offer a reduced price than its competitors. This may be because pricing algorithms are designed to be linked to the prices set by competitors making undercutting guaranteed up to a certain level, or because competitors are resident on platforms themselves. Studies by Harvard Business School and Upstream Commerce revealed that out of 857 tracked items first offered for sale by Amazon Marketplace sellers, within 12 weeks, Amazon began selling 25 percent of their top-selling items. Various factors then place Amazon at an advantage – because of economies of scale, it may be able to offer those goods at a discount to competitors; because of packages like Prime, it may attract buyers more easily than competitors; or by virtue of its name, reliability and speedy delivery, it might attract buyers who don’t have Prime but who know and trust the Amazon brand.

4. Developing new products

Data allows platforms to analyse which products are popular and to develop similar products in their own name. Economies of scale and developed goodwill (this essentially means the same as above – knowing and trusting a brand because of its reputation and name leading to more consumers using that brand) make for an advantage over competitors. This is exactly how the AmazonBasics range was conceived, developed and continues to flourish.

We can take the impacts of giant tech platforms beyond the sociological impacts outlined in the social dilemma and see that these platforms are either driven by advertisers or driven by self-interested platforms looking to develop their own products and/or tailor the products they sell to be attractive and competitively-priced.

Regulatory responses

the social dilemma ends almost emphatically suggesting that regulation is one of, if not the only, viable large-scale solution to these issues and law is already moving in several key areas to respond to data collection and utilisation.

1. Data privacy

Remember when the GDPR was first introduced and we all had one mad week of a million emails telling us that the GDPR had been introduced, we all clicked that we agreed to terms and conditions which we never dreamed of reading, and then life seemed to go back to normal for many of us? We might hear “GDPR” a lot but it’s easy to forget just how significant the legislation is with regard to protecting personal data.

Whilst the GDPR is incredibly significant and important, when it comes to double- and multi-sided platforms, issues with data collection and utilisation are more often than not linked to observed data, not personal data. Amazon doesn’t need to know my name, it wants to watch my browsing habits. Instagram doesn’t need to know my age, it can see how long I look at pictures, adds and what I search for to gauge what advertisements and content I will be receptive to – from which it is likely to have a decent indication of my age. There are exceptions, of course. Geographic location and other aspects of personal information are incredibly useful to platforms wanting to knit together a holistic view of their consumer, but personal data is often used to this end, not to be sold or passed on. All of this means that the GDPR often doesn’t go far enough when it comes to regulating data in the context of double- and multi-sided platforms.

2. Competition law

Competition law aims to ensure competition in markets on the premise that this drives innovation and efficiency, leading to improved, cheaper, more accessible products. Data frustrates competition since large platforms are able to collect information on market trends and react to those trends, improving their business model, products and services to the detriment of smaller businesses. At least as early as 2016, Margarethe Vestager was talking about the role competition law has to play in the regulation of Big Data. Since, the EU has been clear that it wants data to be an imperative aspect of competition law. More recently, EU authorities have presented draft rules which target the utilisation and sharing of data, aspiring to remedy the disparities in access to data in markets and subsequently improve competition.

Killer acquisitions are an obvious example which clearly fit within the competence of competition law. An ex-post assessment of merger control prepared by Lear for the CMA (the relevant competition authority in the UK) suggested that Google, Amazon and Facebook had completed an aggregate of 299 acquisitions from 2008-2018, many of which were not subject to a Tier 1 investigation by the CMA. The approach to merger control is changing and competition authorities are moving to more rigorously scrutinise acquisitions by double- and multi-sided platforms, even where these acquisitions would not traditionally invoke such investigations.

Competition law is increasingly stretching and flexing to reign in double- and multi-sided platforms in their collection and use of not just personal but also observed data. This is likely to be a hot topic in competition regulation and jurisprudence over the next few years as ideas become legislation.

So, whether you realise it or not, your scrolling is being used to fuel business models. Your behaviour online is swept up by platforms who harvest and convert this data into value, through improved services, targeted acquisitions, responsive pricing and stockpiling strategies, and the development of new products. Regulators are hurrying to slow these data factories down, but for now it remains to be seen whether legislation can rebalance competition in markets and give consumers comfort and control over the data they provide and create.