Experimentation · Feedback · Analysis
Describe a change and the AI editor builds it on your live page. Pertento runs the A/B test, collects what your visitors actually think, and reads your pages for what to try next. One platform, and nothing that follows anyone around the internet.
12-day free trial · 0.9 KB loader · Cookie-free by design
AI editor
Pertento's AI editor reads the page, proposes the changes and previews them on your live site, so anyone on the team can build a strong experiment without writing code.
The console
A statistics page tells you the size of the effect, how confident you can be, and what it does not prove yet. In plain words underneath every number.
0.9 KB
No cookies, no personal data.
27%
Across experimentation programmes.
>15%
Runs through Pertento.
GDPR & CCPA
Cookie-free by design.
Solutions
Run it yourself, resell it to your clients, or let us run the whole programme. Same platform underneath in every case.
Self-serve platform
Everything your team needs to run experiments in-house, on the world’s lightest script.
Managed CRO programme
We run your optimisation programme end to end: research, design, build and analysis.
Partner / agency
Manage experimentation across every client from one place, with results a client can read without you translating them.
Cookieless data
Cookie-based platforms only measure the visitors who accept the banner, typically around three quarters. Pertento needs no cookies and no banner, so you capture every visitor. More complete data means tighter confidence and faster decisions.
~75%
Of visitors measured
Around a quarter of your visitors decline tracking, leaving gaps and noise in every experiment.
100%
Of visitors measured
Every visitor counts, so you reach significance sooner and decide with the whole picture.
Why teams move to Pertento
It was trusting the result, keeping a broken experiment off a client’s site, and knowing why the number moved. Pertento is built around those three problems.
Trust the result
Frequentist and Bayesian results sit side by side, sequential testing lets you look early without cheating, and a sample-ratio-mismatch alarm tells you when the data itself is wrong.
Stay across it
One screen ranks every experiment across every account by what it wants from you: measurement that has broken, a result that has reached significance, a collision with another test. The rest keeps running and stays out of your way.
Know why
Surveys run inside the variant, so free-text answers arrive attached to the version the visitor actually saw, themed for you rather than left in a spreadsheet.
0.9 KB loader
Test on all visitors, not a sample.Cookie-free by design
GDPR and CCPA compliant.27% average uplift
Across experimentation programmes.Performance and privacy
Most experimentation platforms ship a script in the hundreds of kilobytes, and a tag manager in front of one adds its own container before the tool inside it can even run. Pertento’s loader is 0.9 KB, loads in the head, and never blocks rendering. The features you switch on add their own code on top of it.
It tracks no personal data: no cookies, no fingerprints, no cross-site identifiers. Compliance is built in, not bolted on.
0.9 KB is the loader; a page also carries the experiments and features you have live, so the total varies by site. Comparison figures are published uncompressed script sizes for ten testing platforms, December 2025, plus a median tag-manager container.
Integrations
One snippet drops into any site, or go fully server-side. Nothing to re-architect.
Paste one tag in your head and you are live: a 0.9 KB loader, no build step, no dependencies.
Deploy and manage experiments straight through GTM. No developer time required.
Run tests beyond the browser: pricing, search and routing, flicker-free.
Not on the list? If it renders HTML, Pertento can test it:read how each one connects.
What’s new
We've been looking forward to this one. The Pertento AI editor is live, and it changes how you build A/B tests. Until now, building a variant meant clicking through elements, hunting for selectors, or waiting for a developer to find time. You had the idea. Getting it onto the page was the slow part.
Introducing Program: a view of your whole experimentation program, not just the tests running today.
Pertento now tells you why visitors behave the way they do, runs your testing roadmap for you, and brings it all together in a completely redesigned platform. Surveys and Pipelines are now in beta
We are excited to announce a new partnership with Kustom, one of the leading checkout and payment providers in the Nordics.
Customers
“With just one small change, we doubled the conversion rate in the shop. I highly recommend this team. They are incredibly professional, efficient and super nice to work with!”
“Pertento uncovered incredible insights into areas we had never even considered, mind-blowing. The results were outstanding; I never thought we’d be able to increase this much.”
“Worked with Pertento on various occasions to increase conversion rates and sales. We are very satisfied. Very professional and helpful, coming up with suggestions after suggestions for improvements and prioritisation. Very much appreciated cooperation!”
“Conversion optimisation is, to start with, a badly underestimated area to bring in help on in ecommerce, and one that can genuinely give a business enormous lifts.”
Pricing
Invite the whole team, and the client. Every feature is on every plan, and each account is priced on the sessions it actually runs.
€390 / month
Start: 100,000 sessions and 2,000 AI credits a month.
Start a 12-day free trial€299 / month
Unlimited client accounts. Each client account is on its own plan as well.
Book a callFrom €1,300 / month
We run the programme. Platform included, quoted on scope.
Book a callQuestions
One script tag in the head, or a package if you deliver from the server. Nothing else is required for experimentation, feedback and analysis, which share the same runtime.
The loader is 0.9 KB, and there is no container in front of it and no second request for feedback or analysis. It loads asynchronously and never blocks rendering. Anti-flicker is not the tag holding the page up: the runtime hides the page while it applies changes and reveals it again, from four independent code paths. The tag still belongs in the head, because one at the foot of the body records exposures just as well but lets visitors watch the original flash into the test.
Yes. The 0.9 KB is the loader, the part every page carries. Each feature you switch on adds its own code to what the page runs, and a live experiment adds the changes it makes, so a large variant weighs more than a small one. What a page loads is the loader plus whatever you have running on it.
Not for the experiment itself. Pertento sets no cookie and stores no identifier of its own: the variant a visitor is given is kept in their browser storage, so what they see stays consistent on later pages and on later visits. That is there to keep their own experience coherent rather than to profile them. Sending the same exposures on to GA4 is a separate decision, and GA4 does need consent.
On Pertento's own tracking, all of them. That is the practical difference between measuring without an identifier and measuring behind a consent gate: a platform that has to store one before it can count a visitor only ever sees the visitors who agreed, and those are not a random sample of your traffic. Report the same experiment through GA4 instead and you are back inside its consent gate, which is why the two will legitimately disagree.
Feedback runs surveys inside the variant, so a free-text answer arrives attached to the version the visitor actually saw rather than to the site in general. Analysis reads the same runtime and the same exposures, so the numbers behind a result and the numbers behind a survey are the same numbers. Both are in production on customer sites and both are included on every plan.
Visitors see the original page. Assignments are cached at the edge and the runtime fails open, so an outage costs you data collection, never a working site.
Yes. The console is built around it: pick an account and a website at the top, or leave both open and the attention queue spans everything you are responsible for.
Edge network
Built and hosted in the EU, with points of presence on every populated continent. Your experiments run close to your visitors wherever they are, which means faster delivery, lower latency and better data.
A new experiment is a draft. It only reaches visitors once you start it and the website is published, so nothing goes live by accident while you are learning the tool.