Digital Regulation & Privacy Retrospective

Did the GDPR cost online shops traffic? What the 2019 research really measured

Early studies found falls in recorded EU web traffic and revenue after the GDPR. We revisit what they measured, what they could not, and the lessons for analytics today.

Key points

  • A July 2019 working paper by Goldberg, Johnson and Shriver, using Adobe Analytics data from 1,508 firms, found recorded EU page views and recorded revenue fell by about 10% after the GDPR enforcement deadline.
  • The peer-reviewed version, published in 2024, reports a 12% fall in both measures for 1,084 firms and separates real effects from changes in data recording.
  • Related work found websites cut their use of third-party technology vendors by 15% in the week after enforcement, with concentration among the largest vendors rising.
  • A drop in analytics numbers after a consent change is not the same as a drop in sales; reconcile analytics against your order system before drawing conclusions.

A year after the GDPR’s enforcement deadline of 25 May 2018, economists began publishing the first hard data on what it had done to European websites. Some of the early headlines suggested online shops were losing visitors and sales. The research itself was more careful, and its central caveat, that the law changed what websites could measure as well as what customers did, is the part most worth remembering.

The July 2019 working paper

The most cited early study was a working paper by Samuel Goldberg, Garrett Johnson and Scott Shriver, dated 16 July 2019 and titled “Regulating Privacy Online: The Early Impact of the GDPR on European Web Traffic & E-Commerce Outcomes”. It used web analytics data supplied by Adobe for 1,508 firms using the Adobe Analytics platform.

The authors used a difference-in-differences design. Because the GDPR applied to all EU traffic, there was no untouched group of EU websites to compare against, so the working paper used each site’s own traffic from the same period of 2017 as the comparison, to account for seasonal patterns. Their abstract reported that recorded page views and recorded revenue fell by about 10% for EU users after the enforcement deadline, and that the change came from fewer visits being recorded rather than shorter visits: time on site and pages per visit stayed broadly constant.

Note the repeated word “recorded”. The paper set out two channels through which the GDPR could reduce the numbers. One was real: a regulation that raises the cost of using personal data can weaken marketing channels such as email and display advertising that bring visitors to a site, and may make some users browse differently. The other was measurement: firms might collect less analytics data, or fewer users might consent to it, so that genuine visits simply stopped appearing in the dashboard.

The published version

The study went through several revisions and was published in February 2024 in the American Economic Journal: Economic Policy (volume 16, issue 1, pages 325–58) under the title “Regulating Privacy Online: An Economic Evaluation of the GDPR”. The final sample covers 1,084 online firms. Its abstract reports a 12% reduction in both EU user page views and website revenue recorded by the platform after the enforcement deadline, and states that the authors decompose these effects into changes in real outcomes and changes in data recording.

The difference between “about 10%” across 1,508 firms in 2019 and 12% across 1,084 firms in 2024 is a reminder that working-paper figures move as samples and methods are refined. Anyone citing this research should cite the published version and use its own description of how much of the effect was real.

Other early evidence

Several other studies looked at the plumbing of the web rather than outcomes.

  • Third-party vendors. In a companion study later published in Management Science in 2023, Johnson, Shriver and Goldberg tracked the technology vendors used by more than 27,000 top websites. Use of web technology vendors fell by 15% for EU residents in the week after enforcement, and because sites were relatively more likely to keep the largest vendors, concentration in the vendor market rose by 17%, with Facebook- and Google-owned vendors gaining share. The aggregate changes faded by the end of 2018, but the effect persisted for advertising vendors.
  • Consent and tracking. Using data from an online travel intermediary, Aridor, Che and Salz (RAND Journal of Economics, 2023) found that the GDPR’s opt-in requirement led to a 12.5% drop in the consumers the intermediary could observe, while those who remained were trackable for longer and became more valuable to advertisers on average.
  • News sites. A Reuters Institute study by Libert, Graves and Nielsen compared news websites in seven countries in April and July 2018. The share of pages with third-party content (99%) and third-party cookies (98%) did not change, but the number of cookies set without user consent fell by 22%.

Taken together, the 2019-era evidence suggested that the GDPR reduced data collection and the use of some tracking vendors, at least initially, and that analytics totals fell. It did not show, on its own, that European consumers bought substantially less online.

Limitations worth keeping in mind

Recorded is not real. When a visitor declines analytics cookies, a consent-dependent tool may never see the visit. A fall in recorded revenue can therefore coexist with stable orders in the shop’s own back office.

Who was in the sample. Firms using an enterprise analytics product are not representative of all online retailers, particularly small shops on hosted platforms.

Timing. The early studies covered the months around May 2018, when enforcement had barely begun and many consent banners were rudimentary. Behaviour, banner design and regulators’ expectations have all changed since.

Two laws, not one. Cookie consent derives mainly from the ePrivacy Directive, which predates the GDPR; the GDPR raised the standard for valid consent and the cost of getting it wrong. Studies attributing effects to “the GDPR” are often measuring the combined effect.

Where things stand (as of November 2026)

The rules that produced these effects remain in force. Consent is still required for most non-essential cookies and similar technologies in the EU, and consent-based measurement gaps are a normal feature of European analytics. Reform is under discussion: the Commission’s Digital Omnibus proposals, which include changes to the cookie rules, were still being negotiated between the EU institutions at the time of writing, and their final form is uncertain. Our guide to cookie consent and analytics for ecommerce tracks the current position.

Our assessment: what it means for ecommerce analytics today

The lasting lesson from the 2019 research is methodological, and it applies to any shop that changes its banner, its analytics tool or its consent settings.

  1. Reconcile before you react. Compare analytics revenue with orders in your shop platform or payment provider. If orders are steady and analytics revenue has fallen, you have a measurement gap, not a sales problem.
  2. Track your consent rate. Know what share of visitors accept analytics, and annotate dashboards when banner design or wording changes.
  3. Understand what your tool estimates. Some analytics products fill gaps with modelling or offer cookieless measurement modes. Check what is observed and what is inferred; our comparison of ecommerce analytics tools sets out the differences.
  4. Measure conversion on consistent data. When calculating conversion rates, make sure numerator and denominator come from the same consented population; using analytics to improve conversion covers this in more depth.
  5. Do not design consent around the numbers. A banner nudging people into acceptance may improve dashboards but creates legal risk; see our guide to GDPR obligations for online businesses.

Bottom line

The 2019 studies are often summarised as “the GDPR cost websites traffic”. A more accurate summary is that websites recorded less traffic and revenue from EU users after May 2018, that part of that drop reflected what could be measured rather than what happened, and that the researchers themselves worked to separate the two. For an online shop, the most useful inheritance is the habit of asking which of the two you are looking at.

This article is general information about past research and current law, not legal advice.

Sources and further reading

  1. Goldberg, Johnson and Shriver, 'Regulating Privacy Online: An Economic Evaluation of the GDPR', American Economic Journal: Economic Policy 16(1), 2024 — American Economic Association
  2. Goldberg, Johnson and Shriver, 'Regulating Privacy Online: The Early Impact of the GDPR on European Web Traffic & E-Commerce Outcomes', working paper, 16 July 2019 — NBER conference paper
  3. Johnson, Shriver and Goldberg, 'Privacy and Market Concentration: Intended and Unintended Consequences of the GDPR', Management Science 69(10), 2023 (abstract) — EconPapers (RePEc)
  4. Aridor, Che and Salz, 'The effect of privacy regulation on the data industry: empirical evidence from GDPR', RAND Journal of Economics 54(4), 2023 (abstract) — IDEAS (RePEc)
  5. Libert, Graves and Nielsen, 'Changes in third-party content on European news websites after GDPR' — Reuters Institute for the Study of Journalism
  6. Regulation (EU) 2016/679 (General Data Protection Regulation) — EUR-Lex
  7. Directive 2002/58/EC (ePrivacy Directive) — EUR-Lex