What can you test with Optimizely Web Experimentation?

Use it for changes that run on a website page. For example, show one group of visitors a short signup headline and another a version that explains who the product is for. The visual editor or custom code can define the versions.

Your site needs the Web Experimentation script on the pages involved. Optimizely Feature Experimentation is the separate product for experiments built into feature logic through a software library.

How do you preview two page versions without affecting visitors?

Before opening the Visual Editor, Preview or any page that loads the experiment script, verify the setup from its settings. The disposable site must load the test project’s script, and the experiment must target that site only. Confirm that signup records, notifications and measurement go only to test destinations. If you cannot verify those connections, compare the two headlines on paper.

Choose a single visible change
Write two versions of the headline in the test project. Leave the form and the rest of the page the same so you can tell which change you are looking at.
Open each variation in Preview
Check both messages on a narrow and a wide screen. Make sure the new text still tells the reader what happens next.
Check the event in the Preview Feed
Configure the outcome event at the point where the test app accepts a signup. In Preview, submit one made-up response and try one the app should reject. The Feed should show the success event only for the accepted signup. Preview activity stays out of live experiment results.

Why might you see the original headline first?

The page can briefly show its original content if the experiment script applies the change after that content appears. Where and when the script loads affects this.

Preview helps you check whether the right variation works and the event fires. Its timing differs from a live run. Evaluate page load behavior separately on an isolated copy with the same loading setup you intend to use.

What should you count before choosing a winner?

Choose an outcome that matches your question. If you want more people to sign up, count accepted signups for each version. A click on the button may only show that someone started.

Check that both groups can record that outcome. If the results still indicate that more visitors are needed, leave the result open. A preview can check that measurement works. It cannot tell you which headline will perform better with visitors.

What should you check in the quote and before moving?

Optimizely directs customers to request pricing. It says cost can depend on traffic, products and how the implementation is set up. Ask which experiment and data options are included for your account.

The documented event export can provide decisions and conversions the service received through cloud storage. Confirm access and retention for your account. Keep the page variations and metric definitions separately so you can explain or reproduce the test later.