How does whoever sends the newsletter know that I opened the email?

You open a newsletter at half past seven in the morning, look at it for three seconds, close it. Somewhere, in a dashboard you will never see, a row has appeared with your address, the date, the time, the type of device and the city you connected from.

The mechanism is an image. One pixel wide, one pixel high, transparent, with a different address for every recipient. When your mail program downloads it in order to show it to you, the server hosting it knows it was you.

In Italy there is a date to mark, and it is close: 29 October 2026.

What actually reaches the sender

Not the contents of your mailbox, not your other messages, not your passwords. The pixel is an HTTP request for an image, and what it delivers is the set of information any request carries with it.

The French data protection authority, on the page with which it presented its own recommendation on 14 April 2026, lists it in usefully concrete terms: whether the message was opened, the date and time of opening, the type of device, approximate location worked out from the IP address at country, region or city level, and how often you read it. And it adds the thing that changes the weight of all of it: this data can be cross-referenced with other information to build a profile. The same authority also states the limit, and it is worth saying because it settles a widespread fear: the pixel gives no access to the contents of your other emails, nor to your mailbox.

Apple, on its legal page in Italian, adds an element that often escapes notice: without protections, the sender can know not only when you open the message, but how many times, and whether you forward it.

Why open rates are not worth much any more

There is a small irony here worth telling: the pixel is still everywhere, but it has almost stopped working as a measurement.

Since 2021 Apple has had Mail privacy protection, and the way it works is radical. Remote content gets downloaded in the background at the moment the message is received, regardless of what you do. The traffic passes through two middlemen run by different entities: the first knows your IP address but not the content, the second the content but not the address. The result: the open is recorded even if you have opened nothing, and the time is not the time you read it.

The sending platforms say so plainly enough in their own documentation. Mailchimp writes that for these contacts open rates come out inflated and imprecise, that it cannot estimate location, device or mail program, that tests based on opens become unreliable, and that resending to people who did not open does not reach anybody with that protection on. Brevo reaches the same conclusion and advises segmenting on clicks instead of opens. Both of them note that clicks are untouched.

Gmail does something too, in a different way. Google’s guide in Italian says images are checked before being received, that this stops senders getting information about your computer or your location through the image loading, and stops them setting or reading cookies in the browser. But the same page also states the limit, and the limit is what counts: senders can sometimes tell whether you have opened an email with images in it. So the geographical data and the device are screened off, the signal that you opened it is not.

Something you often read that I could not verify: the technical description of Google’s system as a proxy with an image cache. The word “proxy” does not appear in Google’s documentation in Italian, and that description circulates on technical blogs from ten years ago. So I am telling you how it works according to Google, not according to the anecdotes.

And something I could not find anywhere, although you see it quoted often: an official percentage for how many opens are false. Neither Apple, nor Mailchimp, nor Brevo publishes numbers. Any figure along the lines of “forty per cent of opens are not real” has no primary source behind it.

The 29 October date

On 17 April 2026 the Garante per la protezione dei dati personali, the Italian data protection authority, adopted its guidelines on the use of tracking pixels in email communications, decision number 284. They were published in the Gazzetta Ufficiale on 29 April 2026, with a six-month period to comply. The Garante’s own summary gives the exact date, 29 October 2026, and adds that compliance will be checked in the inspection activity of the second half of 2026.

What they say, in short. The pixel falls under article 122 of the Italian privacy code, the one that implements in Italy the European rule on access to and storage of information on a user’s device: the same rule that governs cookies, and it is written in a technology-neutral way, which is why it covers an invisible image too. The rule therefore becomes prior consent, freely given, specific and informed, in particular for the individual tracking of opens used to optimise campaigns, decide how often to send, or profile interests.

Some cases stay outside the consent requirement, and they are tightly drawn: aggregate and anonymous statistics on open rates obtained with a pixel that is identical for every recipient of the campaign, meaning one that cannot be traced back to the individual person; authentication and security needs, such as activating an account or resetting a password; mandatory, institutional or service communications, from phishing warnings to changes in contract terms.

Two points concern you directly. The first: withdrawal has to be easy and also selective, meaning you have to be able to say no to the tracking without having to unsubscribe from the newsletter, for example through a link in the footer, and the refusal must not limit your access to the service. The second: the privacy notice has to be clear and given in advance, and for processing already under way it has to be given with the next mailing. There is also a technical requirement I was pleased to find: the identifiers have to be non-sequential and linked to the email address in a separate internal layer, so that the address does not travel inside the pixel request.

A note of transparency about how I worked: the Garante’s pages read badly with the tools I use, and the PDF of the full decision would not open. The summary above is consistent across three Garante documents, the decision, the press release and the plain-language guide, but it is not a word-for-word quotation. If you have to work with this professionally, open the text in the Gazzetta.

At European level the structure rests on a document of the European Data Protection Board, guidelines 2/2023, adopted in final form on 7 October 2024, which deal expressly with pixels in email. They say two things which together close the circle: the arrival of the pixel on the device constitutes storage, even if it is not permanent, and the collection of identifiers constitutes access. From there the rule applies. The same guidelines treat addresses carrying tracking parameters the same way.

France got there first with a similar recommendation, adopted on 12 March 2026 and published on 14 April, with a shorter deadline and limited to addresses collected before publication, and with a practical suggestion I find sensible: do not put the consent request again for six months after a refusal.

Something I looked for and did not find: a fine, Italian or European, specifically about pixels in email. There are two decisions that come close but are about something else, a Norwegian one from June 2025 on pixels in websites and the French fine of 325 million against Google from September 2025, which concerns ads inserted between emails and consent to cookies. Presenting those as fines about the pixel in email would be a mistake.

How to switch it off

Blocked images neutralise the pixel, and the cost is that emails arrive without their graphics until you decide to load them. After a few days most people discover that newsletters read perfectly well like that.

A warning about the names here too. The Gmail ones I verified on the Italian documentation, so I give them as the Italian interface shows them. For the other programs the manufacturer’s documentation is in English, so the routes are the ones given but the exact labels you read on your own screen. And on one specific point I had to stop: whether Apple Mail’s protection is on by default I could not establish, because Apple’s documentation gives clues pointing in opposite directions. The legal page describes the background download as default behaviour, the Mac guide talks about it being turned on the first time the program starts, which implies an explicit choice. Check on your own device.

1. Gmail on a computer: the gear icon, "Visualizza tutte le impostazioni" (see all settings), the "Immagini" (images) section, choose "Chiedi prima di visualizzare immagini esterne" (ask before displaying external images), then "Salva modifiche" (save changes) at the bottom of the page.
2. Gmail on Android: the menu at the top left, Impostazioni (settings), your account, the "Utilizzo dati" (data usage) section, "Immagini", then "Chiedi prima di mostrare" (ask before showing). Yes, the items here are called something different from on the website: that is how it is.
3. Gmail on iPhone: the menu at the top left, Impostazioni, "Preferenze email" (email preferences), "Immagini", then "Chiedi prima di visualizzare immagini esterne".
4. Apple Mail on iPhone: Settings, Apps, Mail, Privacy Protection. The switch for Mail activity protection is there, and if you turn it off the two separate options appear, one to hide your IP address and one to block all remote content.
5. Apple Mail on Mac: Mail, Settings, Privacy, with the same logic: the separate options, among them blocking all remote content, only appear if you take off the overall protection.
6. Classic Outlook on Windows: File, Options, Trust Center, Trust Center Settings, the "Automatic Download" section, and you leave ticked the box that stops pictures in HTML messages being downloaded automatically. The good news here is that the default behaviour is already to block, and Microsoft's Italian page uses the expression "pixel di rilevamento" and points out that they are not the same thing as a read receipt.
7. Thunderbird: Settings, the "Privacy and Security" panel, the section on message content, and you leave the option that allows remote content switched off. Here too blocking is the default behaviour, and there is a button for exceptions if you want to allow a few senders.

For Outlook in the browser and for the new Outlook for Windows I did not find Microsoft documentation giving the route. The official page about the “Block external images” setting covers only the Android and iOS apps. The advice going around, according to which on Outlook on the web for business accounts there is no general setting and you proceed message by message, comes from a forum and not from the documentation. So: I do not know, and if I find out I will add it here.

The “unsubscribe” button

This is worth explaining, because there is a difference between the link at the bottom of the message and the button your mail program shows at the top.

That button at the top rests on a technical standard, RFC 8058 of January 2017, which defines how a sender declares that it accepts unsubscribing with a single click: a header in the message with a secure address, a second header declaring the function, and a valid digital signature covering both. Your program sends a request to that address, without cookies and without credentials, and only with your consent. The standard came about to solve a concrete problem: some mail programs automatically fetched the addresses in headers, causing unintended unsubscribes.

It is not a legal obligation, but it has become a condition for getting mail delivered. Google has required it since 2024 of anyone sending around 5,000 messages or more in 24 hours to personal Gmail accounts, limited to marketing and promotional messages: transactional ones, a booking confirmation for instance, are excluded. A link in the body of the message or an email address is not enough, the header is needed. Senders who already included a link had until 1 June 2024. Google recommends processing requests within 48 hours; Yahoo sets similar requirements and asks that unsubscribes be honoured within two days, and expressly states that it does not want to give a volume threshold.

A distinction the Garante has made explicit and that was not there before: unsubscribing and withdrawing consent to tracking are two different things. From 29 October you have to be able to do the second without doing the first.

To understand what sits underneath

The mechanism by which an innocent-looking address carries information about you is the same one that makes it worth knowing how to read a link before touching it: I talk about that in How to spot a scam text message: the signs of smishing and what to do. And if what interests you is the general picture of who knows what about you while you use your phone, there is Does Google know where I have been? The three switches to check.