Experiments
A/B testing that needs no cookies and no second tool
Build another version of a page, choose a goal and press Start. A share of the page's visitors is sent to the new version, both groups are counted, and you are told when one has won. One experiment at a time is free.
No card required. Set up in 60 seconds.
No cookies, no banner
Which version a visitor gets is worked out from the request each time and never stored. Nothing is saved in the browser, so there is nothing to ask consent for.
Works with the code you have
The tracking code on your pages already runs the test. One optional line in the page's head makes sure nobody glimpses the original first.
A winner you can trust
A winner is only called after the shortest duration and the visits you set, at the confidence you choose. No stopping early on a lucky streak.
Results in plain words
Next to the p-value you get the chance that each version beats the original, and an explanation on every column.
Set up in three steps
Name the page you want to improve and the page with your new version, pick a goal (or create one on the spot), then check each version with its preview link before any visitor sees it.
- Up to 10 versions of a page in one test, with the share of visitors you give each
- Preview links that show a version to you and count nobody
- A prompt for your AI assistant that builds the variant page and adds the script
Results you can act on
Each version is compared with the original on the goal you chose. You see how far the test is from an answer, and you are warned when the visits are not shared out as they should be, the usual sign of a broken variant page.
- Conversion rate, uplift, likely range and chance to beat the original
- More goals and revenue per visit next to the main goal
- Results by device, source and country
What happens when it ends
The experiment ends by itself and the results arrive by email and, on Pro and up, on your webhooks and in Slack, Teams or Discord. If a new version wins, every visitor can be sent to it straight away, so you gain before you have changed your site.
- Steer more visitors to the leading version while the test runs, with a multi-armed bandit (how it works)
- Choose who takes part by device, country or source
- Schedule the start, and keep two tests from sharing a visitor
By plan
Free runs one experiment at a time with the original and one variant. Pro and Business run as many as you like with up to 10 versions each, and add steering, targeting, more goals, breakdowns and scheduling, and can count likely humans only, using Advanced Bot Detection.
Watch: A/B testing without cookies (no consent banner needed) (2:21)
A/B testing without cookies: run a split test on the script already on your site, every visitor in the test, no consent banner, and only real winners.
Transcript
You changed your headline last month. Did it sell more, or did it just feel better? If you don't know, every change you make to your site is a gamble.
I'm James, founder of Privatus Analytics. Let me show you how to stop gambling.
Here's how it usually goes. You rewrite the page. You watch the chart for a week. It looks a bit better. Someone on the team says it worked. Then next month it's back where it was, and nobody knows if the headline did anything at all. So you change something else. And something else. Every one of them a guess. You tried a testing tool once. It needed a cookie to remember who saw what. That meant a consent banner, and everyone who clicked no was left out of the test. The test only counted the people who said yes. It was another subscription too, on top of your analytics, for a question you ask a few times a year. And the time it did show a winner, you called it after three days, on a lucky streak that never held up. So you went back to guessing. And every guess costs you a month of sales you can't get back.
With Privatus Analytics, the test runs on the script that's already on your site. Nothing to install, nothing extra to pay for. Pick the page, add your new version, and say what counts as a win, like a sale. Press Start. No cookies, so no banner, and every visitor is in the test, not just the ones who clicked accept. Then you wait, and the result is one you can trust. In this example, the new checkout converts at 9.3 percent, against 5.4 for the original. That's 72 percent more sales from the same traffic. A winner is only called when it's real. Never after three days on a lucky streak, and never because of bots, which machine learning filters out of every visit. When it wins, every visitor goes to the winning version straight away, before you've touched your site.
That headline you changed last month? Now you'd know, in one number, whether it sold more. Every change becomes a test instead of a gamble. And every visitor counts, with no banner in sight.
Test the change you've been wondering about. Start free at privatusanalytics.com.
Privacy by design
No cookies, no fingerprinting, no profiles. IP addresses are used in memory and never stored.
Start free, upgrade when you grow
Your first workspace gets 14 days of Pro, with no card. After that, stay on Free or pick a plan.
Business
From $49 a month
250k events a month
For agencies, larger teams and security requirements.
Start free in 60 seconds
Add one script tag and see your first visitor within seconds. Free for your first 10k events a month.
No card required. Set up in 60 seconds.