It is the first interaction
Before your headline, before your proof, a visitor meets a box asking them to make a decision about something they don’t care about. That is a genuine cost, and it is usually implemented by whoever installed the tool rather than designed.
The cost is measurable and often ignored
Banners delay content, cause layout shift, and on small screens can occupy most of the viewport — three of the fixed elements that leave no room on a phone. All of that is separate from the consent question itself.
You can’t remove the banner. You can stop it being the slowest, largest thing on the page.
Fewer trackers, smaller problem
Most of the banner’s complexity is proportional to how many third parties you load. Auditing what actually gets used frequently removes half of them, which shrinks the consent problem and speeds up the site simultaneously.
Don’t let it move the page
A banner that appears after render pushes content around at the moment someone starts reading. Reserve its space or overlay it, because layout shift is felt more sharply than load time.
Make the honest choice easy
Dark patterns that hide the reject option are increasingly a compliance risk in themselves and they irritate the exact people evaluating whether to trust you. Equal prominence is both the safer and the better-converting choice.
Remember it applies to your measurement
Declined consent means gaps in your analytics, which is worth knowing before you interpret a drop as a traffic problem. It is another reason the five numbers worth watching should include ones that don’t depend on tracking.