The product wasn't failing — the water was

Roughly one order in fifty came back with the same complaint — the product dried to a spotted finish — and the team could never reproduce it. Their own tests, run in the same building for three years, looked perfect every time.

A review carries no geography, but a return does: plotting return comments and message threads against the ship-to state on their orders made the pattern obvious — they clustered in hard-water regions. The product behaves differently above a certain mineral content, which is a known property of the chemistry and was written nowhere on the page. Customers in those states were not doing anything wrong; they simply needed a different rinse step, and nobody had told them.

We treated the water as a product variable instead of a mystery.

• Complaints joined to the order's ship-to state, which is what turned an unreproducible defect into a measurable condition.
• Hard-water instructions on the page — the extra step stated in the image stack, not buried in the bottle's back label.
• Search coverage for the condition — the brand now ranks for hard-water and spotting terms, which is exactly the shopper most at risk of a bad experience.
• A dedicated variant for the worst regions, so the guidance has a product behind it rather than only advice.

One-star reviews fell from seven per thousand orders to two, and the hero listing's running average recovered from 3.8 to 4.2.

Revenue grew 77% — partly from the rating, partly because the hard-water terms brought in shoppers actively looking for a product that works in their conditions. The repeat rate rose 14 points, which is the real signal: customers who used to conclude the product was bad now conclude it works.