The challenge
In fashion, the products worth promoting can change from one week to the next. New arrivals gain momentum, best-sellers shift and seasonal products quickly move in and out of demand.
Gina Tricot was managing product selection through manually defined segments, such as groups containing the top-performing products by revenue. But static segments require regular maintenance and can quickly become outdated as product performance changes.
Creating separate campaigns and product sets for individual segments also meant splitting budget and conversion data across multiple smaller campaign structures, giving Meta less data to optimise against.
Gina Tricot wanted to improve ROAS, conversion rate and order revenue without increasing spend, while creating a simpler and more scalable way to manage product selection.
The solution
ROI Hunter introduced SMART Segments together with Segment Prioritisation to help Gina Tricot dynamically identify which products deserved greater exposure within its Meta catalog campaigns.
Instead of relying on manually maintained product groups, SMART Segments automatically evaluate products based on signals including performance, conversion behaviour and account benchmarks. As performance changes, products can move between segments automatically.
Segment Prioritization then uses these signals inside a broader product set to influence which products Meta should promote first, prioritizing stronger products while reducing exposure for weaker performers.
This allowed Gina Tricot to maintain a broad catalog while reducing the need to split campaigns into multiple narrow product sets. More conversion data could stay concentrated within fewer campaigns, supporting Meta's optimization while simplifying campaign management.
Gina Tricot ran a controlled A/B test in Germany comparing its existing business-as-usual product selection against a campaign powered by SMART Segments and Segment Prioritization.
Both test cells used the same campaign structure, creatives and virtually identical budgets, making product selection the key difference between the two approaches.
Spend was almost perfectly matched.Performance was then compared across order revenue, ROAS, conversion rate, sessions and revenue per session.
The result
With virtually identical spend and traffic, SMART Segments generated significantly more revenue and stronger efficiency.
🚀 Germany A/B Test
- +17.9% Order Revenue
- +17.9% ROAS
- +19.4% Revenue per Session
- Higher Conversion Rate
- identical spend
The uplift did not come from buying more traffic. The campaign generated more value from approximately the same number of sessions by improving which products were prioritized.
One concern before the test was whether prioritizing products algorithmically could simply increase the share of discounted products and improve performance at the expense of full-price revenue.
The SMART Segments cell did record a slightly higher markdown share — 21.5% compared with 18.9% for BAU. However, the performance uplift remained even when looking beyond markdown-driven revenue.
This showed that stronger results were not simply the result of promoting more discounted products. SMART Segments helped Gina Tricot identify and prioritize products with greater revenue potential across the catalog.
Beyond the performance uplift, SMART Segments gave Gina Tricot a more scalable way to manage product selection.
Instead of repeatedly rebuilding static segments as trends and best-sellers change, product groups can update automatically based on current performance signals. Segment Prioritization then applies those insights without requiring the account to be broken into multiple narrow campaigns.
For a fashion catalog that changes constantly, the result is a campaign structure that can react faster to product performance while requiring less manual maintenance.
By combining SMART Segments with Segment Prioritization, Gina Tricot increased order revenue and ROAS by 17.9% with virtually identical spend — demonstrating how smarter product-level signals can drive stronger results without sacrificing catalog scale.
