The questions section was the second product page, and nobody was writing it

A $400 heater is a considered purchase, and the sequence shoppers follow is consistent: main image, price, reviews, then the customer questions — because that is where the specific, awkward questions live. Does it work with a 20lb tank. How far does the heat actually carry. Will it tip in wind. Can it run under a covered patio.

The brand had answered four percent of them. The rest had been answered by other customers, often confidently and wrong — one top-voted answer stated the heater could not be used under any cover, which is untrue and had been sitting there for two years collecting votes. The section was simultaneously the most-read content on the page, the least controlled, and the reason for a meaningful share of returns: people were buying on incorrect information and discovering it on delivery.

We treated the questions section as owned content with a maintenance schedule.

• Every open question answered by the brand, starting with the highest-voted wrong answers and working down.
• A weekly answering rhythm, so a new question never sits long enough for a stranger's guess to accumulate votes.
• The recurring questions promoted into the page — the twelve asked most often moved into the image stack and A+, which is where they should have been answered in the first place.
• Specification gaps closed — the questions exposed facts missing from the listing entirely, like coverage radius in feet and minimum clearance, which were then stated plainly.

Conversion nearly doubled, from 3.9% to 7.2%, and revenue grew 67%.

The returns halving was the unexpected part: a good share of returns had been caused by the page's own misinformation, and correcting it removed them. Promoting the twelve most-repeated questions into the image stack was the single highest-value change — the same answers had been available all along, buried in a section that required three taps to reach, below a wrong answer with more votes than any of the correct ones.