Overview
Malys Équipements supplies the furniture and equipment of hair salons, barbershops, beauty institutes, spas and medical practices. Its store runs on Odoo, sells worldwide and displays its catalogue in many languages. In other words: a large catalogue, a lot of URLs, and an e-commerce engine that was not designed for search first.
The starting point
A large B2B catalogue served by an e-commerce engine (Odoo) that was not designed first for SEO: the visibility potential was largely untapped, as is often the case in this setup. The entry audit drew up the list of the classic high-volume workstreams.
Page indexability, automatically generated titles and URLs to normalise, load time to reduce, and structured data to add to the catalogue: all levers to fix before producing more pages. The most strategic issue, though, was not in SEO but in the reliability of measurement.
Objectives
- Make the catalogue genuinely indexable, rather than produce more pages
- Make every language work, without sacrificing any
- Give Google and the AIs the structured data of a real catalogue
- Finally know what organic search earns versus advertising
Making revenue measurement reliable
You cannot arbitrate a budget between organic and paid when both share a false revenue figure. And the figure reported in analytics was exactly doubled. "Exactly" is the word that puts you on the trail. A 5% gap is imprecision; a clean factor of two is an addition performed twice.
To settle it, I did the one thing that actually settles it: a real test purchase on the store, paid for, to walk the checkout funnel end to end and inspect what each page was sending. Twelve euros to close weeks of debate.
The diagnosis came back clean: two tag containers coexisted, one still holding advertising tags five years old; Odoo was not feeding the expected data layer, so the order amount was hard-coded into a tag instead of being read properly; and residue from a measurement tool dead for years was still sitting on hundreds of pages. The order was therefore being counted by two mechanisms in parallel.
What that really costs: as long as revenue is wrong, every budget decision is a gamble. You cannot tell whether advertising earns its budget, nor what organic search actually brings in. Fixing measurement is not a nice-to-have technical chore: it is what makes every subsequent decision defensible.
What I did
- A full technical audit across 10,000 crawled pages, with a backlog of 34 prioritised recommendations, and above all defensible one by one, with evidence.
- A semantic study of metadata: 8 parameterised templates to generate titles and descriptions at catalogue scale, rather than writing them one by one across thousands of references.
- Checkout funnel diagnosis via a real test purchase, which identified the double counting and opened the way to fixing it on the Odoo side.
- Structured data: module installed and tested on staging, products and categories finally exposing their price and availability in a machine-readable way.
- Hunting junk addresses: a superfluous parameter was slipping onto nearly 190 of the 195 product links on category pages, multiplying duplicates for Google.
- Multilingual set straight: language tags, translated addresses and per-language metadata corrected.
The decision I did not make
The audit showed that 69% of Google's crawl budget was going into languages other than the main ones. The reflex recommendation, the one you read everywhere, would have been: cut languages. I did not make it.
Because selling worldwide is a deliberate strategic decision by the company, not a technical accident. My job is not to amputate a commercial strategy to make an indicator look better: it is to make what the company decided actually work. Crawl budget is earned through technical quality (translated addresses that respond, clean language tags, translated metadata), not by deleting markets. A consultant who optimises a number by shrinking their client's scope is not helping them.
Results
Over the latest tracked month, impressions are up 28% and clicks up 6%: the catalogue is starting to be seen. The revenue double-counting is identified and its correction under way. Structured data is tested and ready for production. The foundation is being rebuilt in order: make it indexable, make it measurable, and only then push.
What is still outstanding
The work is far from finished, and that is normal on a catalogue this size. Making revenue reliable remains priority number one: until it is closed, you cannot say to the cent what each channel earns. Structured data awaits activation in production. Display performance remains the heaviest piece. And redirects from old addresses progress as we go.
Business impact: a store where three quarters of the pages escaped Google is crawlable again, its languages have stopped tripping over each other, and its catalogue is starting to speak the language of search engines and AIs. Above all, the company will soon know what each euro invested returns, channel by channel, which, on a catalogue this size, is worth more than any SEO trick.