Cookie Banner Copy: The Sentence That Decides the Accept Rate
A small site rewrote its banner copy and nothing else. Same layout, same button size, same colors. The old text said the site cares about privacy and uses cookies to improve the experience. The new text said the site uses analytics to see which pages people read, and advertising to show relevant ads, and gave the visitor a real choice. The accept rate moved ten points. The design did not change. The sentence did.
Banner copy is the most underrated lever in consent, and the most commonly wasted. Most banners say the same thing: a vague sentence about cookies, a link to a privacy policy nobody reads, and two buttons. The visitor learns nothing about what actually happens to their data, so the choice is not informed, and the accept rate reflects that. A banner that names the actual processing gets a different response, because the visitor is deciding about something real.
What the copy actually has to do
The copy has three jobs, and most banners do one of them badly. First, it has to say what data is collected and why, in words a visitor would use. Analytics to see which pages people read is a reason. Improving your experience is not a reason, it is a slogan, and visitors have learned to ignore slogans. Second, it has to make the choice visible: accept, reject, and customize all in the same place, with the reject option as easy to reach as accept. Third, it has to be short enough to read. A banner with four paragraphs of legal text is a banner nobody reads, and a choice nobody reads is not a choice.
The SecurePrivacy guide to A/B testing consent banners puts copy at the top of the list of things worth testing, ahead of button color and placement. That matches what the benchmarks show: ConsentStack's consent rate data tracks accept rates by placement and wording, and the spread between the best and worst wordings is large enough to matter. The sites that test copy find the same thing every time. Specific beats vague, and honest beats friendly.
How to test it without guessing
Test one change at a time, and test it on real traffic. Change the first sentence only, run it for two weeks, and compare the accept rate and the reject rate together. A change that raises the accept rate by hiding the reject button is not a win, it is a dark pattern, and it is the kind of thing that draws complaints. The metric that matters is the informed accept rate: people who read the banner, understood it, and chose to accept. That number is lower than the raw accept rate, and it is the only one worth optimizing.
Watch the reject rate too, because it tells you whether the copy is honest. If the reject rate collapses to near zero after a rewrite, the copy is probably hiding the choice rather than explaining it. A healthy banner has a visible reject rate. The exact number depends on the audience, but a banner nobody rejects is a banner nobody read.
The rewrite that works
Start with the first sentence. Replace the generic line about cookies with a specific one: what you collect, and what you do with it. Then check the buttons. Accept and reject should be the same size, the same color weight, and next to each other. The customize link should be visible without scrolling. Then read the whole banner out loud. If it sounds like a lawyer wrote it, rewrite it until it sounds like a person explaining something to another person.
One last check: the banner has to survive translation. If the site runs in more than one language, the copy has to be rewritten in each language, not machine-translated word for word. A banner that reads naturally in English and awkwardly in German is a banner that German visitors will not trust, and the accept rate will show it.
Keep reading: the dark patterns that mislead visitors, the four signals consent mode v2 tracks, and the mobile bottom-sheet banner cover the nearby cases in this series.
Frequently asked questions
Does banner copy really change the accept rate?
Yes, and the change is large enough to measure. Testing guides and consent rate benchmarks both show that specific copy outperforms vague copy by a wide margin. The sites that test copy find the same pattern: naming the actual processing, analytics and advertising, gets a different response than a generic line about valuing privacy.
What makes banner copy a dark pattern?
Hiding the reject option, making accept bigger or brighter than reject, or writing copy that steers the visitor toward accept without explaining the choice. A banner that raises the accept rate by making rejection hard to find is not better copy. It is a misleading design, and it draws complaints.
How long should I run a banner copy test?
Two weeks on real traffic is a reasonable minimum. Test one change at a time, the first sentence only, and compare the accept rate and the reject rate together. A change that raises acceptances while the reject rate collapses to near zero is hiding the choice, not explaining it.
Should the banner copy be the same in every language?
No. The copy has to be rewritten in each language the site runs in, not machine-translated word for word. A banner that reads naturally in one language and awkwardly in another loses trust with the second audience, and the accept rate shows it.