E-commerce SEO Seminar Case Study: B2B Shop, 12,000 URLs

Around 12,000 URLs in the shop, about a tenth of them in Google's index, three language versions that do not know each other, and more than ten percent of all search queries containing the brand name: That was the starting point of an industrial manufacturer whose marketing and shop team attended my two-day seminar "E-commerce SEO" at GFU Cyrus AG in June 2026. This case study shows which topics really had priority in two days and what the team took away. Company, products and domains are anonymized.

Starting point: B2B manufacturer with a shop, but no SEO process

The company does not sell directly to end customers but through installers who fit the product. Still, end customers see the manufacturer's logo in the end, and some of them later search for spare parts or for products such as ventilation units that have a long purchase decision. So the shop has to serve two audiences, but it is populated by a subsidiary with product data from the PIM, often without maintained translations and without any look at search terms.

Tracking existed: GA4, plus Piwik Pro being set up. SEO experience was limited to meta descriptions and keywords. Nobody had systematically analyzed Search Console before. That is exactly where we started.

Day 1: Foundation, indexing and languages

Who actually searches: installers, end customers, spare parts

The morning belonged to the basics: what the foundation of SEO is, how to find your own place in search, and why AI visibility is relevant but not a foundation when the basics are missing. The most interesting discussion was about whom the shop wants to reach at all. There is search demand for spare parts; for the main products the end customer rather searches via the installer. The competitor analysis with an SEO tool showed: for the target keywords the actual competitors did not appear at all, but dealers and shops for spare parts did. That set the content strategy for day 2.

A projection made the order of magnitude tangible: if the shop ranked first for all observed search terms, that would bring around 160 additional clicks per month, at an average conversion rate of 2.5 percent just a few orders. That is sobering but important: for a B2B shop the lever is not traffic, it is spare parts, availability and brand loyalty.

12,000 URLs and the question of what Google should see

The afternoon was technical. In Search Console, around 12,000 known URLs faced only about 10 percent indexed pages. The causes were typical for shops that have grown over time: variant and filter URLs without noindex, pages with a German title and English text, products without description, PDF files from the technical documentation linked inside the shop. We went through the sitemap index, which lists products, landing pages and categories separately, with the rule that one sitemap may contain at most 50,000 URLs. The team's question about what happens when a new page is marked as index-relevant but is not in any sitemap was the moment the image of the "city map for Google" clicked.

Three languages, missing return links

The biggest technical finding came with the hreflang check: the three language versions referenced each other, but not back. "Missing return link" on every URL we checked. Plus canonical tags pointing to other language versions and English pages whose translation was simply missing because the product had been created at short notice. For Google this means the same page three times under different paths without a clear assignment. The ID card analogy I use to explain canonical and hreflang went with the team into the coordination with the shop agency.

Day 2: Keywords, content and next steps

Brand searches and positions 10 to 20

Day 2 started with the queries. More than ten percent of all impressions came from the brand name, a strong signal of brand loyalty through installers, but also a hint that generic demand is barely captured. The keywords with potential sat in positions 10 to 20: Google sees the pages as relevant, but user signals and content are not enough for page one. Some of them were seasonal topics such as heat recovery, for which the shop runs promotions but has no permanent page. We sorted the keywords into groups: brand, product, local, semantically related and informational.

Category descriptions, FAQs and jargon

Then we moved to content. For a product category without its own text we sketched a category description with the tool and optimized the title, from 82 to 93 points in the content score. The discussion about FAQs was honest: for category pages the team considered them too imprecise because the products have too many special cases, for guide pages they made sense. My note: FAQs do not bring rankings, they answer the questions customers ask Google, and exactly those questions are content ideas.

A topic that only comes up in B2B seminars: jargon. The company uses its own terms for components, which translators faithfully carry over but nobody searches for. The solution is keyword research per language so that product pages contain both the technical term and the search term.

Are AI bots even there?

Finally, a look ahead. The robots.txt did not block any AI crawlers, yet a colleague found not a single AI bot access in the server logs. The explanation from the seminar: anyone with few indexed pages and hardly any structured content will not be visited by AI systems either. The order is therefore clear: indexing and languages first, then content, then GEO.

Learnings for your e-commerce SEO seminar

• Indexing is the first construction site. With a tenth of pages indexed, every content measure is wasted.
• Check hreflang with return links. Three languages without back references are three duplicates.
• Read the brand share. Over ten percent brand searches show brand strength and missing generic visibility at the same time.
• Positions 10 to 20 first. That is where the pages sit that reach page one with little effort.
• Translate jargon, not just language. Keyword research per language belongs in the translation process.
• GEO comes after SEO. Without indexed, structured pages no AI bot shows up.

Conclusion

The team went home with a clear order: clarify indexing and language versions with the shop agency, then build out category texts and spare parts pages for the keywords in positions 10 to 20, then merge jargon and search terms per language. A list that can be worked through in a quarter. For a B2B shop with 12,000 URLs that is worth more than any keyword list.

More reports from seminars are in my workshop case studies, e-commerce projects under e-commerce SEO case studies.