First published on
19 July 2018
Identification
ISBN 978-92-9200-929-8

The aim of this study was to explore the nature and prevalence of online personalised practices (personalised ranking of offers/personalised pricing/targeted advertising), assess whether online retailers/providers are transparent about online personalisation, look into consumers’ awareness and perception of such practices and problems experienced and try to gain insights into the economic effects of online personalisation. It included, inter alia, an EU28-wide online consumer survey, a mystery shopping exercise in 160 e-commerce websites, 8 EU Member States (CZ, DE, ES, FR, PL, RO, SE and the UK) and 4 market sectors (TVs, sport shoes, hotels rooms and airline tickets) and an online behavioural experiment(1) (in the same 8 Member States) which took place in Q2/Q3 2017. 

KEY FINDINGS

  • The study found evidence for ‘personalised ranking of offers’ (websites changing the order of search results when different consumers search for the same products online). Over three fifths (61%) of the e-commerce websites visited for the mystery shopping were found to personalise the ranking, either based on information about the shoppers’ access route to the website (through price comparison websites, search engines or via a mobile device etc.) or based on information about the shoppers’ past online behaviour (e.g. history of visits/clicks).Overall, the share of websites practising personalised ranking of offers was 92% for the airline ticket websites, 76% for hotel room websites, 41% for the websites selling sports shoes, and 36% for the websites selling TVs.
  • The mystery shopping did not find evidence of consistent and systematic personalised pricing(2) (prices being customised for some users for the same products) across the 8 Member States and 4 markets covered. Price differences between personalisation and ‘no personalisation’ scenarios were observed in only 6% of situations with identical products. Where observed, price differences were small, the median difference being less than 1.6%.
  • The proportion of experiment participants correctly identifying personalised ranking of offers increased (from 29% to 40%) as communication transparency by the online platform about this practice increased, for example with explicit information in the purchase page (‘some results are recommended so as to provide you with the most relevant products’) or a text underneath certain (personalised) products (‘based on your previous purchases/history’) in the search results page. However, there was little difference in the proportion of respondents correctly identifying personalised pricing or targeted advertising, as transparency in the communication increased.
  • Survey respondents were most concerned about their personal data being used for purposes other than the ones for which it was gathered and/or not knowing with whom it might be shared (between 36% and 49% for the three practices).
  • A (relative) majority of EU28 online consumers see both benefits and disadvantages of online targeted advertising and personalised ranking (51% and 49% respectively), while 36% of respondents said the same for personalised pricing.
  • Consumers would be more positive about personalisation if they received more information and had more control over these practices. Six in ten EU28 online respondents said they would be more positive if there was an easy option to refuse.

How will these findings be used?

The policy approaches suggested in this study should be seen in the light of the newly applicable General Data Protection Regulation (GDPR) and the reform of the ePrivacy Directive. The study suggests actions to enforce the Regulation’s provisions with respect to online traders’ transparency obligations towards consumers, as well as initiatives to increase the cooperation and information exchange on personalisation practices between consumer and data protection authorities. Self-regulatory actions, such as the development of EU-wide standards and best practices on the use of personalisation practices, are recommended for the e-commerce industry. The study also suggests ways to increase consumer awareness and concrete ideas for further research in this area.

 

(1) This simulated an online search platform where participants were asked to purchase one of eight products listed there, based on information shown to them in the beginning of the experiment about their previous, similar searches. Participants were shown inter alia higher or lower prices (in the personalised pricing scenario), search results based on either their browser or previous search history (in the personalised ranking of offers scenario) or targeted adverts under ranked or not ranked results (in the targeted advertising scenario). 

(2) The mystery shopping findings data on pricing should be interpreted with care. The advanced technological means for online personalisation are extensive and developing rapidly, and hence difficult to detect by a research methodology, especially since pricing algorithms, increasingly used for both price discrimination and dynamic pricing, are often involved. 

Related documents