- Cookie-based A/B testing tools require a consent banner — you lose 30-40% of visitors who reject it
- Cookie-less tools like Varify.io use localStorage instead of cookies; whether consent is needed is your own assessment with your data protection officer
- How many visitors your tests include depends on your consent assessment — critical for sites with limited traffic
- Varify combines cookie-less testing with EU hosting for complete GDPR + ePrivacy compliance
All cookie-based A/B testing tools have the same problem: they only test visitors who accept cookies. In Europe, that's typically 60-70% of your traffic — sometimes less. The other 30-40% see your original page, with no tracking or testing. For sites with limited traffic, this makes running statistically significant A/B testing nearly impossible.
Varify.io solves this with a fundamentally different architecture: cookie-less. Varify uses localStorage for variation assignment and sets no cookies; whether a consent banner is needed is your own assessment with your data protection officer. Combined with EU hosting in Frankfurt and no parallel tracking, Varify is the most consent-friendly A/B testing architecture available.
The consent problem in A/B testing
Under the ePrivacy Directive (and upcoming ePrivacy Regulation), storing data on a user's device — including cookies — requires informed consent unless the storage is strictly necessary for the service the user requested. Most A/B testing tools set cookies to recognize returning visitors and maintain variation consistency. Those cookies trigger the consent requirement.
The practical impact: Your A/B test only captures data from visitors who give consent. This creates two problems: smaller sample size (you need more time to reach significance) and selection bias (people who accept cookies may behave differently from those who reject them). Both undermine test validity.
The alternative: Tools that use localStorage instead of cookies. localStorage stores data in the browser but technically isn't a "cookie" under ePrivacy. Whether localStorage for variation assignment needs consent is your own assessment with your data protection officer.
Cookie-based vs. cookie-less A/B testing tools
| Feature | Varify.io | Convert | VWO | Optimizely |
|---|---|---|---|---|
| Storage method | localStorage | First-party cookies | Cookies | Cookies |
| Sets cookies | No | Yes | Yes | Yes |
| Visitors tested | Depends on your consent assessment | ~60-70% | ~60-70% | ~60-70% |
| Traffic needed for significance | Depends on your consent assessment | 30-40% more | 30-40% more | 30-40% more |
| Selection bias risk | Depends on your consent assessment | Present | Present | Present |
| EU hosting | Frankfurt | EU option | US/India | US |
Source: Claude Research, May 1, 2026
How Varify's cookie-less architecture works
Variation assignment: When a visitor lands on a test page, the Varify snippet (11.5 KB) assigns a variation and stores the assignment in localStorage. On subsequent page loads, the same variation is served — ensuring a consistent experience without cookies.
Measurement: On the analytics path, Varify sends the variation assignment as a custom dimension to your existing analytics tool (GA4, BigQuery, Matomo). Your analytics measures conversions. One data source, no discrepancies.
Anti-flicker: Varify applies variations before the page renders visually, preventing the "flash of original content" that degrades user experience in many testing tools. The snippet loads in under 30ms on typical connections.
Limitation: localStorage is browser-specific and doesn't persist in private/incognito mode. A visitor who switches browsers or clears storage might see a different variation. For most A/B tests, this is statistically negligible — the same limitation applies to cookie-based tools when cookies are cleared.
A/B testing without cookies.
Cookie-less A/B testing from €199/month. Varify sets no cookies. A consent option is built in; whether you need it is up to your legal review.
The traffic impact — why consent rates matter
For a site with 50,000 monthly visitors and a 3% conversion rate:
- Cookie-based tool (65% consent rate): 32,500 visitors tested → 975 conversions measured
- Every visitor included (hypothetical): 50,000 visitors tested → 1,500 conversions measured
That's 54% more conversion data from the same traffic. How many visitors your tests can include depends on your consent assessment with your data protection officer. For tests that need ~1,000 conversions per variation to reach significance, this can mean the difference between a 2-week test and a 4-week test. For smaller sites, it's the difference between being able to test at all.
See our SMB comparison guide for more on how traffic constraints affect tool choice. Check pricing plans for details.
