See which version wins, with real visitors
Experiments let you test a change against your current page and let real visitor behavior decide. Pick a conversion goal, split your traffic, and see the winner with confidence, all aggregate and anonymous.
- A/B, split URL, and multivariate tests
- Use any tracked event as the goal
- See the winner with a confidence level
A · Control
Conversion
4.2%
B · Variation
Conversion
5.8%
The right test for the change you're making
Run an A/B test to compare variations on one page, a split URL test to send a share of visitors to a different page, or a multivariate test to try several changes at once.
- A/B, split URL, and multivariate
- No-code element edits or your own code
- Weighted variants you control
- Target pages, devices, and dates
A/B test
Control vs variations on one page
Split URL test
Send a share to a different URL
Multivariate test
Combine several element changes
Pick a goal and split the traffic
Choose any tracked event as the conversion goal, from form submissions to click-to-call, then decide how much traffic each variation gets. Every visitor gets one consistent experience.
Traffic split
Let the results settle it
Sentinel reads conversions back by variation and shows the lift with a confidence level, so you ship the version that actually performs instead of the one that sounded best.
Testing that removes the guesswork
Three test types
A/B, split URL, and multivariate, for whatever you're changing.
Any conversion goal
Use any tracked event, like a form submit or click-to-call.
Consistent experience
Each visitor sticks to one variation, so the test stays fair.
A clear winner
See the lift and confidence level, not just raw numbers.
Common questions
No. Assignment uses a random visitor identifier and is aggregate and anonymous. A variation assignment is not protected health information, and no patient data is stored.
Find out what your website is costing you
See your accessibility, privacy-risk, and speed signals in one dashboard, ranked by what to fix first. Set up in minutes.