90% of A/B tests produce less than 1.2% effect. After an A/A test revealed our client's store needed 100+ days to reach significance, we built a scoring framework that classifies every experiment idea into three tiers before it burns any runtime.

Running parallel A/B tests on Shopify is harder than it looks. I dug into how Intelligems actually handles template tests, onsite edits, and price tests under the hood, and figured out which ones can safely stack on the same page and which ones can't.