Skip to content
Opens the cited record in a new tab

CNIL fines FNAC EUR 7 million: a recommendation tracker in the app ran after users refused

Personalization is not strictly necessary. The CNIL fined FNAC EUR 7 million because its app kept sending viewed products to a recommendation provider after users said no.

Administrative fine
EUR7,000,000
Regulator
CNIL
Law
French Data Protection Act Article 82
Stage
Fined by the CNIL restricted committee 22 October 2026 (SAN-2026-015); injunction with daily penalty; appeal not established. CNIL decision

The problem

Acting on a complaint from the noyb (None Of Your Business) association, the CNIL inspected the FNAC mobile app in March 2024. It found that while people used the app, requests went through an API to a third-party domain carrying an identifier FNAC assigned to the user and information about their activity, such as the products they had viewed. The tracker powered personalized recommendations (complementary and similar products) and trained the recommendation algorithm. Users had not consented to it, and the app did not respect users' choice to refuse tracking of their browsing (CNIL, 29 October 2026).

What the CNIL found

IssueFinding
ExemptionRecommendations and algorithm training are not strictly necessary for a service the user expressly requested, and do not enable electronic communication, so the Article 82 exemptions did not apply
ConsentPrior consent was required; the tracker ran without it and despite users' refusal in the app
Fine factorsThe refusal was not respected, the app has a large user base, and there is consistent, well-established case law on personalization and commercial-targeting trackers
FineEUR 7 million: EUR 6 million against FNAC Darty Participations et Services and EUR 1 million against FNAC Direct
OrderShow within one month of notification that the new tool obtains consent where needed, or pay EUR 10,000 per day of delay
PublicityDecision made public given the seriousness of the breach, the scope of processing and the number of people affected

Technical issue

  • Data: a FNAC-assigned user identifier plus in-app activity such as products viewed.
  • Recipient: a third-party domain reached through an API call from the app.
  • Purpose: personalized product recommendations and training the recommendation algorithm.
  • Consent state: sent without consent and despite a refusal.

Inth's technical interpretation: server-to-server style API calls from an app are trackers under Article 82 just like web cookies and pixels. Route recommendation and personalization SDK or API calls through the same consent check as advertising, and keep the refusal state on the device so it survives app restarts.

Illustrative pattern, not the defendant's code. Generic consent API, not a specific c15t call.
// Recommendation API called on every product view, regardless of consent
onProductView(p => recoApi.send({ userId, viewed: p.id }));

// Personalization waits for consent; refusal sends nothing
onProductView(p => {
  if (consent.has('personalization')) {
    recoApi.send({ userId, viewed: p.id });
  }
});

Timeline

DateEvent
Not statednoyb files a complaint with the CNIL
March 2024CNIL inspects the FNAC mobile app
22 October 2026Restricted committee fines FNAC Darty Participations et Services and FNAC Direct EUR 7 million and issues an injunction (SAN-2026-015)
29 October 2026CNIL publishes the decision
One month after notificationDeadline to show compliance, or EUR 10,000 per day

Outcome

The CNIL imposed EUR 7 million in fines and ordered the companies to demonstrate, within one month of notification, that their new tool obtains consent where necessary, on pain of EUR 10,000 per day (CNIL, 29 October 2026). Whether FNAC will appeal to the Conseil d'État was not established.

What it cost

ItemAmountSource
Fine, FNAC Darty Participations et ServicesEUR 6 millionCNIL
Fine, FNAC DirectEUR 1 millionCNIL
Daily penalty if the order is not metEUR 10,000 per dayCNIL

Why it matters

  • Apps are in scope. The CNIL applied Article 82 to an API call inside a mobile app, not only to web cookies.
  • Personalization needs consent. Recommendations and model training are not strictly necessary, even for a retailer's own catalog.
  • A refusal has to stop the flow. Ignoring users' refusal was a stated reason for the size of the fine. Compare Orange and CNIL v. Amazon.

How Inth catches this

  • Consent banner

    Holds personalization and recommendation purposes behind consent in web and app flows, and keeps a refusal in effect across sessions.

  • Website audit

    Loads web pages in a clean profile and records third-party requests before a choice and after Reject All. It does not inspect native app traffic.

Out of reach

  • Inth's audit covers websites, not native mobile apps, which is where the CNIL found this tracker.

Consent records with choice, purpose, country and policy version.

Start a free scan

Checks for your own systems

Your ticks stay on this page. Inth does not save or send them.

0 of 6 done

In the repository

0 of 2 done

In the browser and network

0 of 2 done

With the people responsible

0 of 2 done

Sources and limits

The CNIL announcement was opened. The full deliberation SAN-2026-015 on Légifrance and the noyb complaint were not opened. The third-party recipient is not named in the announcement.

Source review: 30 October 2026. This page describes the cited record and procedural stage. It does not audit the company's current systems. Inth wrote the technical checks; they cannot decide which laws apply or prove compliance.