The maintenance spiral
A UI refactor lands, forty tests go red, and none of them found a bug. The selectors just aged out. Teams burn hours re-pointing locators, and after enough cycles someone proposes deleting the suite. Test maintenance, not test writing, is where automation initiatives go to die.
Propose, review, learn
When a selector breaks, our engine works through 15+ fallback strategies (attributes, structure, text anchors, visual position) to find the element the test meant. But it does not silently rewrite the test. The proposed heal lands in a review queue; a human approves or rejects it in seconds; the central element repository records the outcome and gets smarter about that element.
That one design choice, propose instead of overwrite, is the difference between healing you trust and healing you audit nervously. Silent auto-repair fails catastrophically exactly once, and then nobody trusts a green build again.
Measure it for your team
Count the hours your team spent on selector fixes last month. Most teams land between "a day" and "a week." That number, times twelve, is the annual cost the heal-review workflow attacks. Cut it by even half and the platform has paid for itself before you touch any other feature.
How we measured this
- Every heal proposal, approval and rejection is recorded as an audit event, and the review queue is the only path that changes a locator. We do not measure the hours you save; count your own selector-fix time as the article suggests.