The usual habit is to install a tag once, forget it and read the reports. In Europe, that tag sets off two separate rules. The first concerns the visitor's device: storing or reading information on it requires consent, except where it is strictly necessary. The second concerns the data collected: an identifier, an IP address, a behaviour across pages are personal data, with all that follows. Many exporters meet the first when a banner is added by their agency, and never meet the second.
What requires consent, and why
Measuring an audience is not strictly necessary in the legal sense. The visitor asked for a page and the page works without the measurement. That is why a script that counts visits sits on the same side of the line as an advertising pixel, unless a specific national exemption applies. The data side is separate. If the tool's provider receives the data on servers outside the Union, that transfer needs its own legal mechanism, and the position in this area has changed more than once. If the provider uses the data for its own purposes, you have a third party in your visitor's browsing, and you need to be able to say who it is and what it does.
- Nothing non-essential loads before the choice. No analytics script, no advertising pixel, no third-party font, no embedded video player until the visitor has decided.
- Refusing is as easy as accepting. Same level, same number of clicks, comparable prominence.
- No advertising identifier hidden in « analytics ». Check what the tag actually sends, not what the settings page says.
- A record of the choice. What was chosen, when, under which version of the banner text.
- Information in the visitor's language. Who receives the data, for what, and how to change the decision later from any page.
Measuring without a cookie
A family of tools counts visits without placing a persistent identifier on the device. They aggregate page views, sources, countries and device types, do not follow a person across sites, do not keep the full IP address and are hosted inside the Union by a provider that does not reuse the data. Some national authorities accept that strictly statistical measurement of this kind may be exempt from consent, under narrow conditions: a limited purpose, no cross-site tracking, short retention, no reuse. That exemption is not uniform across countries, so read the guidance of the country where you sell. Your own web server logs are another source: they record requests without any script on the visitor's side, though the IP address in them is still personal data and needs a retention rule.
| Method | What it tells you | Where it stops |
|---|---|---|
| Analytics tag with consent | Paths, sources and repeat visits, for those who accept. | Blind to everyone who refuses. The sample is not the audience. |
| Cookie-free statistical tool | Pages seen, sources, countries and devices, in aggregate. | No individual journeys, no retargeting. |
| Server logs | Every request, including from visitors who refuse anything. | Robots mixed in with people. Needs filtering, and shows nothing about behaviour on the page. |
| Enquiries counted on the server | How many enquiries arrived, and from which page. | Only the people who write. |
| A question on the contact form | How the person found you, in their own words. | Depends on memory and goodwill, but often the most useful answer you will get. |
What you lose when you refuse to measure
Without any measurement, the site is run on opinion. You do not know which pages produce enquiries, whether the pages in a European language are reached at all, or which source brings serious contacts: a fair, a directory, a search, a referral. You cannot judge a change, because there is nothing to compare before and after. You cannot tell whether an advertisement did anything. There is also a distortion to keep in mind. With a consent banner, the visitors who accept are not a typical sample, and their share varies by country. That is a reason to measure outcomes, such as enquiries received, rather than to rely on visitor counts.
What you lose by measuring badly
The opposite failure is common too. An agency adds an analytics tag, an advertising pixel, a behaviour-recording widget, a chat tool and an embedded video, and all of them load before any choice. It is visible to anyone who opens the browser's developer tools, which takes a minute, and that is how a competitor, a consumer association or an authority notices. A banner supplied by the site theme that is not actually connected to the scripts is worse than none, because it is a statement that is false. Scripts from a platform hosted in China loaded on a European page add the transfer question on top.
Deciding what you need to know
- List the questions first: how many enquiries, from which pages, from which source, from which country. Start with questions, not with a tool.
- Match each question to the lightest method that answers it. Many need no script at all.
- Inventory every script and third-party resource the site loads, using the browser's developer tools on a clean session.
- Remove what no question needs. Each item you keep is something you have to describe and defend.
- Connect the banner so that non-essential scripts really wait for the choice, then test it by refusing and watching what still loads.
- Write the tracker policy and the privacy notice to match what is actually running, and update them when you add something.
Count outcomes before visitors. Start with a simple record of each enquiry received: date, country, the page or the event it came from, and the answer to « how did you hear about us? ». It needs no script, no consent and no tool, and it is the figure that decides whether the site works. Everything else is explanation, and you can add it later with a method that matches the question.