From pilot to 12 markets: how Gomez scales product segmentation
200k+ products per market, segments built on GA4, Google Ads and Meta Ads data, and feeds for Google, Meta, TikTok, ChatGPT Ads and TradeDoubler.
- markets in under 5 months
- 0
- products per market
- 0k+
- premium and luxury brands
- 0+
- channels from one system
- 0
About the client
Gomez is an online fashion house with physical stores, selling clothing, footwear and accessories from 150+ premium and luxury brands. Every new market, channel and collection adds decisions, so segmentation had to scale fast, repeatably and without manual work.
- Industry
- Premium and luxury fashion
- Catalogue
- 200k+ products, 150+ brands
- Markets
- 12 European countries
- Channels
- Google, Meta, TikTok, ChatGPT Ads, TradeDoubler
Challenges
- 01
Improve sales efficiency with segmentation based on real performance data — from single products to brands and categories.
- 02
Combine feed, GA4, Google Ads and Meta Ads data in one place, separately for each market.
- 03
Extend segmentation with custom labels calculated automatically from results.
- 04
Scale the approach from Poland to further countries quickly.
- 05
Support new channels and feed formats without the store's IT team.
Connected data sources
- Product feed
- GA4
- Google Ads
- Meta Ads
Markets step by step
- May 20265PLDEROHUCZ
- July 20269SKGRNLBE
- September/October 202612BGHRIT
How we did it
Launch in 5 markets
- Poland, Germany, Romania, Hungary, Czechia.
- Each market has its own data model, sources and label set.
Expansion to 9 markets
- Slovakia, Greece, Netherlands, Belgium.
- Segmentation reflects local demand instead of copying Polish settings.
12 markets
- Bulgaria, Croatia, Italy.
- Data is pulled daily; rules place products in the right campaigns and settings.
Custom label architecture
- CL0: zombie / zombie with potential / budget burnersChannels: Meta, TikTokLogic
Item group ID, 30 days: Google Ads impressions (up to 100 or more) versus GA4 add-to-carts.
What it deliversSeparates under-exposed products with real interest from those that spend budget without response.
- CL1: new / new without revenueChannels: Google Ads, TikTokLogic
Products in store for under 30 days, highlighting those without revenue.
What it deliversNew items don't compete for budget with proven products — the autumn collection's visibility clearly improved.
- CL1: revenue segmentsChannels: Google Ads, TikTokLogic
60-day revenue (high / mid / low / no revenue) by item group ID.
What it deliversCampaign structure built around real product value; the 60-day window fits a premium offer.
- CL1: meta potential starsChannels: MetaLogic
Under 30 Meta impressions with at least 1 GA4 add-to-cart, 30 days.
What it deliversCross-channel logic: catches products with interest that Meta isn't showing yet.
- CL2: revenue segments by variantChannels: Google Ads, MetaLogic
60-day revenue by product variant.
What it deliversFiner control over exposure of specific variants in a broad fashion catalogue.
One system, many channels
- Google Ads and Meta Ads
Labelled feeds for each of the 12 markets.
- TikTok
Since August 2026, including a handbags feed that fits the 100 MB limit.
- ChatGPT Ads
Feed ready the same day the channel became available.
- TradeDoubler
Feeds for PL, HU and RO excluding selected seasons and collections.
- Custom feeds
E.g. a list of selected products for specific campaign activities.
Results
- Scale without extra resources
12 markets, 200k+ products and 5 channels run by a compact performance team.
Segments are a daily campaign optimisation tool across many markets.
Thanks to a separate new-arrivals layer, August sales of the new collection beat July.
New feeds and label changes usually within one to a few working days, without IT.
What's next
- 01
Include the Omnibus price in feeds.
- 02
Flag size end-of-lines and older collections based on stock data.
- 03
Automatic labels for products strong in Google but under-exposed in Meta (and vice versa).
“Data Octopus lets us segment products automatically and use the data directly in campaigns across 12 markets. We manage the exposure of new arrivals and high-potential products more efficiently. New feeds and changes go live fast, without involving IT or extra manual work.”
More case studies
- BeautyFull story
GA4+AdsCombined dataHow Perfumeria.pl unlocked the potential of thousands of productsFeed segmentation based on margin, priorities, Ceneo price attractiveness and performance broke through the Google Ads revenue ceiling.Read the story Perfumeria.pl - AutomotiveFull story
4Data sources in segmentationUnitrailer: segmentation that fuels campaign growth in FranceFeed, GA4, Google Ads and margin data in a new product split and campaign structure — results after 7 weeks.Read the story Unitrailer - FootwearFull story
1125%ROAS of activated productsButosklep: 1,929 products given a new chance to sellActivating non-selling products in a separate Google Ads campaign at 1125% ROAS — results after 12 weeks.Read the story Butosklep
Questions and answers
How fast can product segmentation roll out across many markets?
Gomez went from 5 to 12 markets in under 5 months. Each market got its own data model and custom labels instead of a copy of Polish settings.
Which custom labels work for fashion e-commerce?
Gomez uses zombie-with-potential, new-without-revenue, 60-day revenue segments and cross-channel “meta potential stars” labels, among others.
Can one data model feed Google, Meta, TikTok and ChatGPT Ads?
Yes — the same data models produce feeds for every channel, with limits and formats tailored to each.
Your products have untapped potential too
We'll show you which products in your campaigns are waiting to be activated.