Larger updates, fixes and changes to the a/b testing

[marcusljf]

This is a major push to make the A/B testing really functional and useful for people. Defining the new logic and helping clarify a few unknowns.

2-way sync between the destination url and the url in position 1

It was saving from the destination URL to the testing URL, but not the other way around. This is now synced for two-way.

Reorganization of opened link view

When the A/B test is open in the link view, the traffic percentage has been moved to the right for better alignment with the stats.

A/B test tooltip

When hovering the A/B test icon in the links view, we now show that the A/B test is running and provide the completion date.

Completion date visual warning

If a user selects a completion date further out than six weeks, the note changes to a red color to indicate that it's outside that window.

Appending http to urls

When adding testing urls, if HTTPS:// isn't added when focus isn't on the input, it adds it in, much like the destination URL.

End test selection

If the user is ending the test early, we now show the stats when they're selecting the URL as the winner.

Ending modal overflow

If the link overflows to ellipses, and the user can't see the entire URL that they added.

Winner logic updates

• Select the link with the highest amount of conversions
• If conversions are tied, select the link with the highest conversion rate (conversions ÷ clicks)
• If still tied, select the link with the highest clicks

OR - If no conversions are recorded or tracked

• Select the link with the highest number of leads
• If leads are tied, select the link with the highest lead rate (leads ÷ clicks)
• If still tied, select the link with the highest clicks

Other minor updates

• All tool tips have been linked up to the recently published help doc.
• Reduce padding between the traffic split label and the input.
• Small content updates to the ending modal.

Summary by CodeRabbit

• New Features

  • Deterministic A/B winner selection using conversions with a multi-step tie-breaker cascade; no winner if still tied.
  • End-of-test modal shows per-URL analytics with overflow-friendly tooltips and selection flow.

• Improvements

  • Per-field URL inputs normalize on blur, validate, and sync with variants; prettier URL display and dynamic completion-date styling.
  • Analytics and UI consistently use URL normalization.
  • Analytics/preview links use encoded query parameters.

• Removed

  • Legacy A/B testing modal and its prior UI/controls.