Why fivebar sets no cookies
1 November 2026
If your analytics needs a cookie banner, it can only count the people who click Accept. Everyone else quietly drops out of your figures, and there are more of them than you might think.
In the EU, a site has to ask before it stores a cookie for analytics. The UK relaxed this in February 2026 for simple analytics, as long as visitors are told and can opt out. Wherever a site does ask, its analytics has to wait for an answer. Click Reject, close the banner or just leave, and as far as the stats are concerned, you were never there.
Who goes missing
Quite a lot of people, as it turns out. Orbit Media compared Google Analytics with a cookieless tool on its own site. Google Analytics missed 55.6% of the visitors who were shown a banner, against 15.8% of those who weren’t.
A lot depends on the banner itself. In a study of more than 80,000 visitors to a German website, only 2.9% did anything at all with a banner across the top of the page. Researchers at MIT, UCL and Aarhus found that tucking the reject button away on a second screen raised consent by 22 to 23 percentage points. They also found only 11.8% of banners on the UK’s top sites met the law’s minimum requirements. So the fairer your banner, the fewer people you count.
France saw this happen. When its regulator made sites offer a proper Reject button in April 2021, consent fell by 15 points on computers, to 55%, and by 4 points on phones, to 71%.
The people who say no aren’t a random slice, either. In France, 41% of over-35s turned cookies down in 2022, against 34% of under-35s. In a survey of 1,035 adults by OptOut Advertising and Motivaction, highly educated people said no 41% of the time, against 33% overall. And visitors in Germany and France reject cookies most often, while those in the US accept them most. If you sell to careful, well-informed people at their desks, they’re the ones you’re least likely to see.
Filling the gap with guesses
Some tools try to fill in the blanks. Google Analytics’ consent mode can estimate the people who said no from the ones who said yes. But it only does this on busy sites, with at least 1,000 people a day saying yes and 1,000 events a day from people who said no, for at least a week.
On a smaller site, the missing visitors just stay missing. On a bigger one, part of your figures is a guess, and it assumes the people who said no behave like the people who said yes. Going by the research above, they often don’t.
Does the banner put people off?
It’s tempting to say yes, but the honest answer is: a bit, sometimes. A study of 1,084 firms found EU sites recorded about 12% fewer pageviews and 12% less online revenue after GDPR came in. The authors showed part of that was simply less being recorded, rather than fewer people turning up. A larger study of 6,387 websites found visits down 10% after 18 months. Its authors put that down more to sites finding it harder to reach new people than to banners getting in the way.
Where a banner does seem to cost something is when it blocks the page. One sports nutrition shop swapped a banner you had to click through for a strip along the bottom. In the UK, sales conversions went up 32% and revenue per visitor 5%. That’s one shop over two weeks, but it’s what you’d expect.
Banners can slow pages down, too. Google’s own guidance calls them a very common source of layout shifts. On a phone the banner can be the biggest thing on screen, which makes it what Google times as the page loading.
How we count without asking
fivebar sets no cookies and keeps nothing in your visitors’ browsers, so there’s nothing to ask. To tell one visitor from another, we make a one-way code from their IP address and browser, your site, and a secret. At midnight we throw the secret away. So we can tell that two visits today were the same person, but never that someone today was here yesterday. Nobody can, including us.
That does have a cost, and we’d rather be upfront about it. We count each visitor once a day, so over a week your visitors are visitor-days, not people. We can’t tell you how many came back. And someone who moves from coffee shop Wi-Fi to their phone’s network mid-afternoon counts twice.
We think that’s a fair trade. We’d rather count everyone a little roughly than know a lot about only the people who said yes. Content blockers can still stop our script, as they can any analytics, but serving it from your own domain can help.
One more thing: we’re not lawyers. If your site has ads or embedded videos, it’s still worth checking whether those need a banner of their own.
If you’re curious how many people your banner has been hiding, try running fivebar next to your current tool for a week. Our free trial lasts 30 days and doesn’t need a card, so there’s plenty of time. The two will never match exactly, as every tool treats bots and blockers its own way, but the gap will give you a good idea of how many people said no.