Visual editor
Click any element on your own site to change its text or style, hide it, or insert HTML. No browser extension.
A/B testing, hosted in Germany
Change a headline, a button or a whole page for some of your visitors and see which version converts better. TryFast tells you in plain words when a result is safe to act on, and your AI agent can set up the test for you.
Add the snippet to your site, or let your AI agent do it, and connect your cookie banner with one more line. We show you when the snippet loads and when the first visitor is counted.
<script src="https://cdn.tryfast.eu/s/acme7k2pqx.js" async></script>DetectedClick and edit in the visual editor on your own site, write CSS and JavaScript, or describe the change to your agent. Preview every variation before anyone sees it.
#hero h1Hide.trust-rowSet style.btn-buyTryFast splits your visitors, counts conversions and tells you what to do: ship it, keep the original, or keep collecting.
No account needed. Save your workspace to an account whenever you like.
TryFast works with MCP, the open standard AI agents use to work with tools. Connect once and ask in plain language: your agent looks at the page, writes the variation, sends you preview links and asks before anything goes live.
claude mcp add --transport http tryfast https://tryfast.eu/mcpTest a free shipping headline on acme-shop.de/pricing against the current one. Goal: more people starting checkout.
The headline is #hero h1, "Simple plans for every shop". I created the draft "Free shipping headline": Variation B reads "Free shipping on every order". Goal: Started checkout. Preview B on your site: acme-shop.de/pricing?tf_preview=... At about 4,500 visitors a day, detecting a 10% lift takes about four weeks; a larger effect shows sooner.
Ready to launch. Approve it here: tryfast.eu/app/approve/apr_8f2kQ7xLm1
Click any element on your own site to change its text or style, hide it, or insert HTML. No browser extension.
CSS and JavaScript per variation, checked when you save, with the exact line of any error. On every plan.
Send some visitors to a different page. Query strings stay, both sides are counted the same way.
Pages, devices, new or returning visitors, UTM parameters, cookies and your own visitor attributes.
Clicks, page visits, custom events and revenue. Duplicate orders count once.
Share a link that shows one variation on your live site, on any device, without touching results.
Traffic split, tracking and mixing checks every hour. A broken split blocks the ship verdict.
React, Next.js, Vue and others: route changes are picked up without extra code.
Every action through a documented REST API, and a log of who changed what, people and agents alike.
Most A/B testing statistics are only right if you look once, at the planned end. People look every morning, and AI agents look every hour; each look raises the chance of crowning a winner that is not one. TryFast uses sequential statistics by default: the intervals account for every look, so a result marked safe to act stays right.
Servers in Nuremberg, backups in Germany. Every subprocessor is listed publicly.
Operated by subprocess.eu/company, Osloer Str. 83, 13359 Berlin, under German and EU law.
By default TryFast waits for your cookie banner and stores nothing before consent. One line connects Cookiebot, Usercentrics, OneTrust and others.
We count visitors with a random id, never with their IP address.
A single first-party cookie, within the 13-month limit EU regulators recommend.
A data processing agreement under Art. 28 GDPR is part of our terms.
Privacy policyData processing agreementSubprocessorsCompliance
Compressed engine. Your experiments add only their own changes.
Engine and experiments arrive together, no second round trip before your page shows.
Launches, pauses and stops reach everyone within two minutes.
Variation code is checked and built ahead of time.
While experiments load, the page is hidden for at most 800 ms by default, and only on pages with an experiment.
We checked public pricing pages, documentation and live scripts in October 2026. Where others do more, we say so.
| Feature | TryFast | Kameleoon | VWO | Optimizely |
|---|---|---|---|---|
| Published prices | Free, €49 and €249 per month | From $495 per month (Starter) | On request | On request, annual contracts |
| Free plan | 25,000 tested users a month | 30-day trial | None listed | None |
| Start without an account | Yes | No | No | No |
| Where data is hosted | Germany | EU or US | US, EU (Belgium) or India | EU hosting available |
| Default statistics | Sequential, valid at any time | Frequentist; sequential and Bayesian available | Bayesian | Sequential |
| Code editor | Every plan | Paid add-on on Starter | Included | Included |
| Visual editor | Text, styles, hide, remove, insert HTML | Full graphic editor | Full visual editor | Full visual editor |
| AI agents (MCP) | Builds, previews and launches web experiments; launches need your approval by default | Remote server for experiments and flags | Local server for feature flags only | Remote server; web variation code is edited in the visual editor only |
| Script your visitors load | Under 12 KB engine, compressed | 49 to 183 KB compressed on five live sites we measured | Not measured by us | Not measured by us |
| Personalization, feature flags, multivariate tests | Not yet | Yes | Yes | Yes |
Checked 24 October 2026 from kameleoon.com, wingify.com/pricing, optimizely.com and each vendor's MCP documentation. Script sizes are brotli-compressed file sizes of the per-site script on five public Kameleoon customer sites. Tell us at support@tryfast.eu if something has changed.
2 projects · 25,000 tested users a month · every feature
10 projects · 250,000 tested users
Unlimited projects · 2 million tested users · priority support
A dedicated server and your own limits
Yes. The free plan has every feature, 2 projects and 25,000 tested users a month, with no card and no time limit. You pay only when you need more projects or traffic.
No. Enter your website and you get a workspace right away. It is kept in your browser; save it to an account any time to keep it for good and invite your team.
The script is under 12 KB compressed and arrives in one request with your experiments inside. On pages with an experiment, the page waits at most 800 ms by default so visitors never see the original flash before the variation.
In Germany and most EU countries, yes: A/B testing is not covered by the exemptions for strictly necessary cookies. TryFast waits for your banner by default and stores nothing before consent. Our privacy docs explain each option.
TryFast has an MCP server at https://tryfast.eu/mcp. Add it to Claude, ChatGPT, Cursor or VS Code and your agent can install the snippet, look at your pages, build experiments, share preview links and read results.
Not by default. When you connect an agent you choose its permission. With the default, the agent prepares everything and sends you an approval link; nothing goes live until you click it. Pausing is always allowed.
TryFast uses sequential statistics. The intervals are built to stay valid no matter how often you or your agent look, so checking daily does not create false winners.
Yes. Route changes are detected automatically and experiments are applied again on every new page. For custom rendering you can call tryfast.refresh().
On Hetzner servers in Nuremberg, Germany, with backups in Germany, operated by a company in Berlin. Visitor IP addresses are not stored. The subprocessors page lists every provider involved.
Yes. The concepts carry over one to one: experiments, variations, goals and targeting. There is no automatic import yet; most teams rebuild their running tests in an afternoon, or ask their AI agent to do it from the old code.
It keeps working while you use it. After 30 days without use it pauses: running experiments pause and your snippet stops changing your site. Save it to an account and you can resume them where they left off. After 90 days without use it is deleted.
Paste one line, make one change, and let your visitors decide.