description

Description is read after title and before action. Few users read it word-for-word; most skim. That makes the first 160 characters disproportionately important — they're the visible portion in compressed placements and search snippets. The remaining body still matters for relevance scoring but rarely directly drives clicks.

The structure that converts: open with the most distinctive product attribute (material, build, key feature), follow with 2-3 specific spec points (dimensions, capacity, runtime), close with a use-case or fit note. Avoid abstract benefits ("amazing quality", "perfect gift") in favour of concrete attributes ("100% merino wool", "30-hour battery", "fits chest 38-42 inches"). Concrete content reads as honest; abstract benefits read as marketing copy and get discounted by both classifiers and users.

The length sweet spot is 500-800 characters for most categories. Under 200 characters triggers "thin content" warnings — the listing serves but with reduced relevance scoring. Beyond 1000 characters you're past diminishing returns and just adding noise that suppresses relevance scoring. The exception is highly technical categories (electronics with specs, beauty with ingredients, supplements with disclosures) where longer descriptions add real information.

The HTML problem catches almost every Shopify/WooCommerce catalog at some point. Descriptions stored as HTML for storefront rendering get exported to the feed as HTML. Channels accept plain text only — embedded tags either display raw in some surfaces (which looks broken) or get stripped along with line breaks (which produces wall-of-text descriptions). The fix is at the feed-generation step: strip HTML, convert <br> and </p> to actual newlines, preserve paragraph structure as plain-text breaks. Most feed tools do this automatically; some pipelines miss it.

The promotional-language ban applies to description as much as to title. "On sale now", "limited time", "best price guaranteed", contact info, links, calls-to-action like "Buy now!" — all get flagged. Description is product description, not advertising copy. Advertising copy belongs in ad extensions or rendered creative; descriptions should read like product specification.

For multi-language catalogs, descriptions need to be in the destination market's language. An English description on a Spanish catalog is treated as low-quality content; channels' content quality scoring downgrades the listing. AI translation works for getting baseline coverage; manual editing of translated content (especially marketing-flavoured passages) catches the issues machine translation leaves behind. The cost of having native-quality descriptions across markets is meaningful but the conversion difference vs machine-translated descriptions is large.

For SEO benefit beyond the catalog channel itself, descriptions feed into AI search results increasingly. ChatGPT Shopping, Google's AI Overviews, Perplexity Shop all extract product information from descriptions to answer shopping questions. Items with detailed descriptions get cited; items with thin descriptions don't. The same content investment serves multiple discovery surfaces — write descriptions well once and they earn traffic across channels.