Building a Product Page for French Cotton Bedding: 6 Things That Moved Conversions
I run a small bedding store, Cozy Home Textiles (https://www.webcozyhometextiles.com/). We sell French-style cotton duvet sets - ruffled edges, lace trim, quilted shams - to customers in the US and EU. There is no engineering team, just a storefront, a spreadsheet of SKUs, and a lot of trial and error.
This post is about the boring part: the product page. Not the design. The structure underneath it.
1. The size table is the whole page
Our first product pages listed sizes the way our supplier did: 200 x 230 cm. Conversion was bad and support tickets were worse. Half the emails were "will this fit a queen?"
We now print both units on every variant:
- Full: duvet cover 200 x 230 cm (79 x 91 in), bed sheet 240 x 245 cm (94 x 96 in)
- Queen: duvet cover 200 x 230 cm (79 x 91 in), bed sheet 265 x 245 cm (104 x 96 in)
- King: duvet cover 220 x 240 cm (87 x 94 in), bed sheet 265 x 245 cm (104 x 96 in)
Same numbers, different framing. Returns dropped. It is not a clever growth hack, it is just doing the reading for the customer.
2. Generate the inches, do not hand-type them
Hand-typed conversions drift. Someone converts 245 cm to 96 in on one page and 97 in on another, and then the size chart contradicts the variant picker.
We generate them:
CM_PER_INCH = 2.54 def to_inches(cm: int) -> int: return round(cm / CM_PER_INCH) Rounding to the nearest inch is fine for bedding. Keeping it in one function means every page agrees with every other page, and the size filter in our navigation can reuse the same numbers.
3. Compare-at price: one number, one rule
We set OldPrice = Price * 1.25 so the struck-through price is consistent across all nine colour/size variants of a set. When we tried to write those fields with a script, they looked correct in the browser and then silently did not persist - the admin form needed real keystrokes. Lesson: verify writes by reading the public page, not by trusting the admin toast.
4. One product, one canonical URL
We had product links with tracking parameters (?Sort=, ?searchword=) getting crawled, and Search Console started reporting a pile of "blocked due to access forbidden" URLs. The pages were fine; the bot protection was rejecting the parameterised variants by user agent.
The fix was boring infrastructure:
User-agent: * Disallow: /*?*Sort= Disallow: /*?*searchword= If you run a storefront, take ten minutes and read your own robots.txt as Googlebot. You will find crawl budget leaking into URLs that should never be indexed.
5. Sitemap freshness is not automatic
New products did not show up in Google for days. The sitemap was being served from a cached build - our platform claimed a 24 hour refresh, and the file on the edge was four days old. URL is unknown to Google in URL Inspection is the tell.
Now we rebuild the sitemap manually the same day we publish a product, then resubmit. Two minutes, and the new URL goes from unknown to indexed.
6. Write the copy like a person describing a bed
The listing copy that converts for us is not keyword-stuffed. It says what the thing is: cotton face, ruffle width, whether the shams are quilted, whether the duvet cover has a hidden button placket. Then it says what it is not: not sateen, not a hotel-white set, not a print that needs a matching wallpaper.
We still keep a title under 60 characters so it does not get truncated in search, and a description around 150-160 so the snippet reads as a sentence. That is the entire "SEO strategy".
The set in the photos above is our French Floral Cotton Bedding Set - cotton, small wildflower print on a cream ground, ruffled hem, two quilted pillowcases:
https://www.webcozyhometextiles.com/products/french-floral-cotton-bedding-set-cozy-home-textiles
Full collection: https://www.webcozyhometextiles.com/
If you build storefronts for a living, the pattern is probably familiar: the wins are not in the design system, they are in the size table, the canonical URL and the freshness of a file nobody looks at.