VARIANTS
16,200
product URLs
Every size and frame had its own product URL, canonical and copy of the description. For one design, Search Console showed impressions spread over 14 variant URLs; none ranked above page three.
Separation 1 of 4
A fictional business. The figures are a modelled scenario, not a client result.
UZOR PRINTS · ART-PRINT SHOP FOR 30 ILLUSTRATORS · PLOVDIV
One page per print, then +123% organic orders
12 months · Bulgaria · Commerce SEO plan
Organic orders / month
+123%
Organic orders / month, month 12 against month 0
01THE CHALLENGE
Each of 900 prints lived on 18 separate URLs, 6 sizes × 3 frames, with heavy images and no product schema, so the variants competed and none ranked.
Uzor sells limited-run giclée and screen prints for 30 Bulgarian illustrators. The shop platform created a separate product for every size-and-frame combination, so a single design existed 18 times: the same title, the same description, a different URL and six full-size mock-up photos each. Google saw thousands of near-duplicates, picked a different variant for each query and each week, and spread clicks and links so thinly that no version reached page one. Shoppers who did land had to load another heavy page just to compare A3 with A4.
THE VARIANT MATHS · ONE DESIGN
1 design
6 sizes × 3 frames
18 URLs per design
× 900 designs = 16,200 product URLs, each with a self-referencing canonical.
02DIAGNOSIS
VARIANTS
16,200
product URLs
Every size and frame had its own product URL, canonical and copy of the description. For one design, Search Console showed impressions spread over 14 variant URLs; none ranked above page three.
Separation 1 of 4
IMAGES
2.4
MB per page
Each variant page loaded six full-resolution JPEG mock-ups, 2.4 MB on average. Mobile LCP at p75 was 5.8 s and 71% of product URLs failed Core Web Vitals.
Separation 2 of 4
SCHEMA
0
product markup
No Product, Offer or ProductGroup markup anywhere: no price or stock in results, and Merchant Center rejected 38% of the feed for prices that did not match the page.
Separation 3 of 4
FACETS
41,000
parameter URLs
Illustrator, colour, size and room filters combined into about 41,000 crawlable parameter URLs. Googlebot spent 58% of its requests there, and new designs took five weeks to be indexed.
Separation 4 of 4
03THE STRATEGY
Built on our Commerce SEO plan: ecommerce SEO for catalogues where templates, variants and feeds decide what ranks.
One URL per design. The 17 extra variant URLs of each print were 301-redirected to the product page, where size and frame became options. A ?variant= parameter keeps the choice shareable and canonicalises to the product. 15,300 redirects, mapped from the catalogue export and tested on staging.
ProductGroup markup with variesBy size and material, and one Product per variant in hasVariant, each with its own SKU, price and availability. The Merchant Center feed is generated from the same data, so the price in results always matches the page.
Six 400 KB JPEG mock-ups became one AVIF hero with srcset and explicit width and height, the rest lazy-loaded. Switching variant swaps a 60 KB image instead of loading a new page: images per page fell from 2.4 MB to 210 KB.
186 illustrator, theme and room collections kept as indexable landing pages with their own copy. Every other filter combination noindexed and kept out of crawl: 41,000 parameter URLs stopped competing with the product pages.
04THE VARIANT PRESS
Here is one design, Kukeri No. 7, as Uzor used to sell it: 6 sizes × 3 frames, 18 URLs. Load each variant onto the press and see what consolidation does to the page, the schema and the images.
The rack · 18 URLs
Drag a variant onto the press, or press it to load. Press again to take it off.
Modelled for one design: positions, shares and weights are illustrative.
Nothing on the press yet
Every variant still sits on its own URL, with its own six photos and no product schema.
Drop variants here
05MONTH BY MONTH
MONTHS 1–2
Full crawl, Search Console impressions by URL, Merchant Center diagnostics, image weights and 12 weeks of logs. Revenue mapped to every design.
MONTHS 3–4
The variant matrix rebuilt as options on one product template. 15,300 redirects mapped from the catalogue export, rehearsed on staging, released illustrator by illustrator.
MONTHS 4–5
ProductGroup and Offer markup generated from the catalogue; the Merchant Center feed rebuilt from the same source, so price and stock always match.
MONTHS 5–7
An AVIF pipeline with srcset, width and height on every image, mock-ups lazy-loaded, and a variant picker that swaps one image instead of loading a page.
MONTHS 7–9
Illustrator, theme and room collections kept as indexable pages with their own copy; every other filter combination noindexed and kept out of crawl.
MONTHS 10–12
New designs indexed in days, illustrator pages linked from every print, gift guides built on the consolidated product pages.
06RESULTS
Organic orders / month
+123%
310 → 690 orders / month
Indexable product URLs
16,200 → 900
15,300 variant URLs redirected to their product
Mobile LCP (p75)
5.8 s → 1.9 s
Field data, mobile, p75
Designs with product rich results
0 → 862
862 of 900 designs, month 12
Modelled scenario for a fictional business — not a client result. Rankings and traffic are never guaranteed; see our Terms.
Core Web Vitals, field data
The tick marks the “good” threshold.
Product page weight
| Query | Competing URLs, before | Best position, before | Month 12 |
|---|---|---|---|
| постер кукери | 14 → 1 | #31 | #3 |
| арт принт пловдив | 9 → 1 | #18 | #1 |
| плакати български илюстратори | 11 → 1 | #44 | #5 |
| kukeri poster | 12 → 1 | #39 | #4 |
| рамкиран постер а3 | 7 → 1 | #62 | #9 |
| bulgarian folk art print | 10 → 1 | #57 | #8 |
Time to index a new design: 35 days → 3 days. Crawlable filter URLs: 41,000 → 186 collection pages.
Selling the same product in many sizes, colours or finishes? Start with the Commerce SEO plan, or tell us how your catalogue is built.