GTIN and Brand: Two Attributes That Change ROAS
A small trick that in practice delivers 20–30% lower CPC. Why most catalogs have GTIN and brand incorrectly filled — and how to fix it.

Most catalogs we receive for audit have one of two errors with GTIN: either it's not filled at all, or it's filled "creatively" — with an internal SKU number that has nothing to do with the Global Trade Item Number standard. Both errors Google treats the same: it ignores your product when a user searches for something generic like "Adidas Run 350".
Rule 1: GTIN Must Be a Real GTIN
GTIN is an 8, 12, 13, or 14-digit number that uniquely identifies a product globally. For products without a GTIN (custom, handmade, white-label) there is a procedure — you set the identifier_exists attribute to false. Everything else is shooting yourself in the foot.
Rule 2: Brand Must Be Accurate, Not an Approximation
"Adidas Originals" and "ADIDAS" and "adidas originals" are three different brands for Google AI. Consistency is more important than accuracy — if you consistently write "ADIDAS", the algorithm will map it to the right brand. But if you use three variations in the same feed, AI treats you as three small brands, not one big one.
In one audit we saw 12 different variations of the brand "Nike" in the same feed of 600 SKUs. CPC was 31% above the benchmark for the same category.
What to Do If Your Feed Is Already in Chaos
- Audit first, not fixing. Run a complete inventory of errors. Without the full picture, you'll be fixing individual SKUs endlessly.
- Bulk fix by pattern. If you have 600 SKUs with 12 variations of "Nike" — that's 12 regex rules, not 600 manual edits.
- Validate before push. Merchant Center has a feed validation API. Use it before every upload.
In practice, for a fashion client with 850 SKUs, these two rules together delivered -24% average CPC and +18% impressions for the same search. Without additional budget. Without changing the strategy.
Do you recognize yourself in this?
Request a free Feed Audit. No commitment, results in 48h.
Follow this blog for upcoming case studies and analyses.
Want this kind of result on your store?
Request a free analysis