Skip to main content
Privacy8 min read

What a consent banner costs your analytics, in visits and revenue

Analytics that waits for consent only counts the visitors who accept. What that does to your traffic, your sources, your conversion rate and your revenue attribution, and what changes when nothing is written to the device.

Vicente TorresFounder of Vero

A consent banner asks one question, and every visitor who says no disappears from any analytics tool that waits for the answer. That is the banner’s real cost to you. Not the pixels it takes or the click it demands, but the traffic, the sources and the sales that never reach your reports. This post walks through what a cookie-based tool actually counts, why the people who decline are not a random slice of your audience, and what the gap does to conversion rate and revenue attribution. Then it covers what changes when the analytics script writes nothing to the device at all, what Vero leaves out, and the one question this post does not answer for you.

Analytics that stores an identifier in a cookie needs permission to put it there, and in many places it has to ask the visitor first (for the EU, see the EDPB guidelines on the ePrivacy Directive). So the script waits behind the banner. Until the visitor accepts, nothing loads and nothing is sent. Decline, close the tab or ignore the banner and keep scrolling, and as far as the report is concerned the visit never happened.

The report does not show a gap. It shows a smaller number that looks perfectly clean, with charts that rise and fall and sources that rank neatly. There is no row labelled “declined”. Every metric you read, from visitors to conversion rate to revenue by source, is computed on the subset who clicked accept and presented as if it were the whole site.

Your server logs and your payment provider do not wait for consent. That is where the gap becomes visible: more requests in the logs than visits in the report, and more sales in your billing than conversions in your analytics. The difference is not noise. It is the traffic the banner filtered out before your tool could see it.

Why are the people who decline not a random sample?

If every kind of visitor declined at the same rate, the loss would be a scaling problem. Your totals would be low, but the ratios would hold, and you could correct them in your head. Nothing guarantees that. Who declines depends on who your audience is, where they browse from and how the banner is designed.

Think about who is most likely to click reject: people who know what a tracking cookie is, people who already run privacy tools, people on a work laptop with a strict policy, people who arrived with a clear intention and want the banner out of the way. If you sell software to developers or other technical buyers, that is not the edge of your audience. It can be the middle of it.

A skewed loss is worse than a large one, because it bends comparisons. It does not take the same share from every source, page or campaign. It takes more from some than from others, and it does so without a trace. A report built on the people who accepted describes the people who accepted. Whether it also describes your customers is a question it cannot answer.

What the gap does to your sources

Picture two channels that bring you the same number of visitors. One is a developer newsletter, the other a broad social campaign. If the newsletter’s readers decline more often, the report shows the newsletter bringing fewer visitors, fewer signups and less revenue than it really does. The campaign looks stronger by comparison, not because it performs better, but because its audience clicks accept.

Budget follows the report. The channel that looks weaker gets less spend and less attention, and the one that looks stronger gets more. Over a few months the banner has moved money from the audience that knows you best toward the audience that is easiest to measure. Nobody decided that. It is what a filtered dataset does to anyone who trusts it.

Direct traffic gets distorted too. A visitor who declines on the first visit and accepts on a later one is counted from the later visit only. The source that first brought them is lost, and the return visit often lands as direct. Direct grows, the real referrer shrinks, and the question of where your buyers come from gets a vaguer answer every month.

Conversion rate is a ratio, so it breaks in two places. The visits underneath only include people who accepted. The conversions on top only include people who accepted and then bought while the script was allowed to run. If declining and buying are related at all, in either direction, the rate you read is not your site’s conversion rate. It is the conversion rate of the people who agreed to be measured.

Revenue attribution takes the hardest hit. Your payment provider records every sale, but a sale can only be credited to a source if the visit that led to it was tracked. Every buyer who declined arrives in your billing with no source attached. In your analytics that money either does not exist or sits in a bucket with no channel, so the report meant to tell you what pays undercounts exactly what you most need to know.

That is the cost worth measuring. Pageviews lost behind a banner are an inconvenience. Sales that arrive without a source are decisions made on the wrong evidence: a campaign cut while it was paying, a page rewritten while it was converting, a channel scaled because it was easy to see.

Put a number on your own gap

We will not give you a rejection rate. It varies with your audience, your visitors’ countries, the banner’s design and the default you choose, and we have no source we would stand behind for a single figure. What you can do is estimate your own, and see what it is worth in money.

The cookie tax calculator takes four inputs: monthly visits, conversion rate, average order value and the share of visits your measurement loses. It returns the visits you never see, the conversions left without a source and the revenue left unattributed, per month and per year. The loss slider starts at 20%, which is a starting value, not a benchmark. To replace it with yours, compare the visits your cookie-based tool records with your server logs over the same week.

A hypothetical example, using the calculator’s starting values: 50,000 visits a month, a 2% conversion rate and a €60 average order. If 20% of those visits never reach the report, that is 10,000 unseen visits, 200 conversions with no source and €12,000 a month, or €144,000 a year, in sales your analytics cannot credit to any channel. Your numbers will differ. The shape will not: a percentage of traffic turns into a sum of money very quickly.

What changes when nothing is written to the device

The consent question is about storing or reading something on the visitor’s device. Vero’s tracker does neither. It sets no cookies and uses no localStorage, so it has no identifier to ask about and no answer to wait for. It loads with the page and counts the visit, whatever the visitor does with a banner your site may show for other tools, unless you choose to gate it behind that banner yourself.

To count unique visitors without storing anything, Vero computes an anonymous identifier at the edge: a SHA-256 hash of the IP, the user agent, the date and the site, salted with a server secret. The IP is used only in memory to compute that hash and is never stored. Because the date is part of the input, the hash rotates daily: the same person produces a different one tomorrow, and yesterday’s cannot be linked to today’s.

That trade is worth stating plainly. Within a day, Vero counts visitors, sessions and the source that brought each one. Across days, a returning visitor counts as a new one, because nothing persists that could recognise them. For revenue, the checkout module adds the visit to your Stripe Payment Links and Lemon Squeezy checkout links, so the sale carries its source to the payment instead of relying on the device to remember it. The core with that module weighs 1.4 KB gzipped, and revenue by source is part of Pro.

What Vero leaves out, and what it still misses

The list is short on purpose: no cookies, no localStorage, no stored IP. Nothing is written to the visitor’s device and nothing stored identifies anyone. The transparency page lists every field the tracker sends, so you can check the claim against the payload instead of taking it from this post.

Cookie-free does not mean nothing is lost. An ad blocker can still stop a browser script from loading, Vero’s included, and a blocked script sends nothing. That leak has a different fix: counting page loads from your edge middleware or your server, where there is no script to block. How to measure past ad blockers explains the setup.

The tracker also stays silent, on purpose, for visitors whose browser sends Do Not Track or Global Privacy Control. That is a choice we made, and it is a gap you should expect between Vero and your server logs. Those are visitors who asked not to be tracked, not visitors lost to a banner.

This post is about measurement, not law. Whether your site needs a consent banner depends on everything your site runs, from ad pixels and chat widgets to embedded videos, and on the rules that apply to you. Vero cannot answer that for you, and neither can any analytics vendor writing about its own product.

What Vero can do is give your legal team facts to work from: what the tracker writes to the device (nothing), what it sends, what it stores and what it never keeps. The facts for your legal team puts all of that in plain language, so they can reach their own conclusion.

Unsure of a term? Read the glossary

Vicente Torres
Vero is built by one person in Spain. I read every waitlist entry and pick the beta accounts myself.
Vicente Torres, Founder of Vero LinkedIn
Hand-picked beta, waitlist open

Ask for a place in the beta.

Anyone can join the list. The accounts we pick get Pro free for the whole beta.

By joining you agree to receive launch updates and accept our Privacy Policy.

See the demo