We want to hear from you: Respond to our call for information, and tell us about harmful practices that potentially breach consumer or competition law.
We all like to find the things we want online quickly and easily. Whether that is the next film to watch, article to read or food to order, algorithms can provide us with individualised recommendations that are relevant and save us time.
But what happens if an algorithm is changing what we see and manipulating the choices we make in ways we don’t know about or understand?
As the UK’s competition and consumer authority, we want to find out where algorithms are harming consumers or reducing competition, and work with companies to address this.
We have therefore launched a new programme of work in the Data, Technology and Analytics (DaTA) unit focused on analysing algorithms.
The first output of the programme is our research paper, Algorithms: how they can reduce competition and harm consumers.
How can consumers be harmed by algorithms?
Say you are looking to book a hotel room for a few nights away. You go to a hotel booking site and enter the number of nights and the type of room you want. The search returns a list of options, which you think are ranked based on the best rates available to you.
But maybe not.
In 2019, our investigation into hotel booking sites found that some sites ranked search results based on the amount of commission hotels paid to the site. They did not disclose this to customers.
Following the investigation, the companies involved in the CMA investigation made commitments to be transparent about these practices.
Consumers are often unaware that they are being targeted by an algorithm, and this makes it harder for them to challenge practices or be aware when harm might happen. In the example above, the harm involved a ranking algorithm.
Some firms use algorithms to set prices that are personalised to a particular person based on their behaviour or characteristics.
For instance, the FCA found that some home and motor insurance firms were using opaque pricing techniques to identify which customers were most likely to renew with them. They increased prices to these customers at renewal each year, meaning some were paying very high prices. This can lead to consumer harm if the techniques used to personalise prices are opaque.
How can algorithms reduce competition?
Many of us use a search engine to look up products we want to buy. Often, we will look for a comparison website that lets us compare products and prices from online retailers. When presented with so many search results, we tend to click the links on the first few results pages.
Previously consumers might have noticed that when searching for products, the first few pages of search results only contained those from Google Shopping, which is Google’s own comparison shopping service. The European Commission found in 2017 that Google was positioning and displaying its own Google Shopping service much more prominently in its general search results. It was subjecting rival comparison shopping services to its generic search algorithms, which demoted these rivals in the search result rankings and made them much less visible to consumers, but exempted its own comparison shopping service from the same algorithms.
Changes to search and ranking algorithms can also unintentionally exclude businesses that rely on them. During our market study into Online Platforms and Digital Advertising, it was reported to us that unexpected changes to Google Search and the Facebook News Feed left some publishers with a reduction in website traffic. This had direct financial consequences for their businesses, with high costs incurred for understanding and optimising their content to adapt to these changes.
Algorithms can also reduce competition by facilitating collusion in three ways:
- They can be used to automatically detect and respond to price deviations by competitors, which could make explicit collusion between firms more stable, as there is less incentive for those involved to cheat or defect from the cartel.
- Firms can also use the same algorithmic system to set prices, for example by using the same third-party software, through which they could exchange information.
- There are also concerns that algorithms can learn to collude tacitly, without firms explicitly communicating with each other.
These can all lead to sustained higher prices.
What the CMA can do
If we suspect that algorithms are being used in harmful ways, what are the various things we could do?
The CMA has established its Data, Technology and Analytics (DaTA) unit, and has built up capabilities in data science, data engineering, digital forensics, behavioural and technology insights. We monitor complaints, news stories, and research papers to identify potential leads on algorithmic harms that affect consumers and competition.
Once we have identified a potential harm, there are several things we can do.
Among other things, we can use a range of techniques to investigate how widespread or problematic the harm is. We can also open a formal investigation where we have reasonable grounds to suspect that a business’s use of algorithms may have infringed consumer or competition law.
Or we can launch a market study or market investigation. These tools allow us to use our information gathering powers to request information, such as data, code and documentation, to allow for more in-depth analysis and auditing. We can then work with businesses to remedy the harm.
Our work programme on Analysing Algorithms supports our Digital Markets Strategy, as well as our work to establish a Digital Markets Unit within the CMA to promote greater competition and innovation in digital markets.
We intend to work closely with the ICO and Ofcom through the Digital Regulation Cooperation Forum in taking this work forward.
We want to hear from you
Do you have specific examples that we should investigate further, where you think they might be harmful or potentially breaching consumer and competition law?
How feasible or effective do you think the different regulatory approaches and actions we have outlined are to identify, prevent and remedy harm?
Leave a comment