First published on
01 October 2018
Identification
ISBN 978-92-9200-945-8

The purpose of this exploratory study was to assess advertising and marketing in social media with respect to the Unfair Commercial Practices Directive, and to identify possible remedies for any problematic practices identified. The study included online discussions with social media users in four countries and behavioural experiments in six, as well as a mapping of of social media providers in Europe, a stakeholder survey and a business-to-business mystery shopping.

Key findings

Provider landscape: Facebook receives 43.6% of browser visits to social media in the 28 EU Member States plus Norway and Iceland; YouTube receives 28.6%. Other leading providers are Twitter (4.3%) and Instagram (2.5%). Together, they account for 80% of the total (2016 figures, excluding mobile apps).

Native advertising refers to commercial content that mimics content generated by non-commercial users, required by law to show a label disclosing its commercial intent. Participants in our behavioural experiments were shown a social media site imitating either Facebook or Twitter, displaying ad formats, and disclosure labels currently in use.

  • One third (33-36%) of experiment participants did not recognise the native ads shown as commercial. In addition, experiment participants were not better able to identify native ads as commercial when the disclosure label was present than when it was absent. More visible disclosure labels increased identification of ads by 13% in Finland, where a very explicit “Advertised” label is used, not at all in Bulgaria where an ambiguous “Promoted” is used.

Social proof indicators such as likes, shares, followers or views communicate the popularity of content to social media users.

  • Summary indicators, such as the number of Facebook “likes” on a native advertisement, can be artificially boosted. However, ad and product evaluation, purchase intention and choice of experiment participants shown a native Facebook ad were not influenced by whether no likes, 10, 100 or 100 000 likes were shown next to the native ad.
  • On Facebook, the fact that a user has “liked” a specific product is sometimes shown next to another product. In the experiments, 66% of participants wrongly believed that the friend’s like referred to the product advertised next to it, which impacted their responses to the ad and product.

Practices forbidden under the UCPD Blacklist such as false limited, expiring or free offers, and false prize winning competitions are common on social media. During the mystery shopping exercise, such types of clearly misleading advertisements were mostly approved by the in-house advertising platforms of social media providers; compliance checks related mainly to the format.

Data-related practices: in the course of the mystery shopping exercises, advertising intermediaries recommended two specific features of Facebook, also mentioned by stakeholders: custom audiences, where advertisers provide contact information to target their customers; and lookalike audiences, where Facebook targets users with profiles that are statistically similar to the advertiser's customers. Most of the social media an intermediaries also insisted on implementing a pixel to track user behaviour on the advertisers' website. 

Policy follow-up

This exploratory study is part of the Commission’s continuous monitoring of the platform economy. The findings will be presented to the national enforcers (the Consumer Protection Cooperation network - CPC), who will decide wheter enforcement action is needed to ensure compliance with relevant consumer law. They may also consider expanding the scope of their ongoing joint action on social media.

The Commission will discuss whether further research is needed with the Member States and stakeholders, together with the possible need for systematic monitoring of social media contexts. The Commission may consider evaluating the impact of the GDPR on the profiling practices of social media during the evaluation and review of the GDPR due by May 2020.

RELEVANT DOCUMENTS