River Island
Predictive product scoring on top of Performance Max: instead of static hero/zombie bucketing, a model assesses a product's potential by neighborhood effect (comparable attributes) to give visibility to new arrivals and long-tail items that the algorithm overlooks for lack of history
River Island raised revenue by 34% and ROAS by 19% on Kidswear by adding to Google Performance Max a predictive product scoring (smec SmartScoreAI) that assesses new arrivals by neighborhood effect rather than by their sales history alone.
Key points
- Predictive product scoring added on top of Google Performance Max for new arrivals and the long tail.
- smec Campaign Orchestrator stack (SmartScoreAI, Dynamic Segments) on a Merchant Center feed.
- Revenue +34% and ROAS +19% on the Kidswear category.
- Evidence level B, confirmed status, finalist at the European Search Awards 2026.
Objective
Surface the new arrivals and long-tail products that Performance Max ignores for lack of historical data, without sacrificing profitability.
The deployment
River Island replaced its static product sorting (hero/zombie buckets) with a dynamic, AI-driven approach on top of Performance Max, through smec's Campaign Orchestrator. Three building blocks: SmartScoreAI assesses each product through a neighborhood effect, comparing it to similar items (type, brand, price, seasonality) rather than to its sales history alone; Dynamic Segments automatically groups products into strategic categories (high potential, safe bets, new arrivals, low priority) that shift with the signals; budget and tROAS orchestration steers investment at the account level and redirects spend toward the best-performing campaigns. On the Kidswear category, smec reports revenue up 34% and ROAS up 19%. The work earned smec a finalist spot at the European Search Awards 2026 in the Best Use of AI in PPC category, with River Island.
Results Proof B
Case study from a specialist partner (smec), quantified and named (River Island, Elvis Mugera) - an interested source. The project is corroborated by its selection as a finalist at the European Search Awards 2026, a third-party program, but the figures are not confirmed in financial results.
How it works
Documented architectureThe stack in detail
- plateforme smec Campaign Orchestrator Orchestration layer on top of Performance Max: budget and tROAS steering at the account level.
- outil smec SmartScoreAI Predictive scoring of product potential by neighborhood effect (type, brand, price, seasonality), which rates new arrivals with no history.
- outil smec Dynamic Segments Dynamic product segmentation (high potential, safe bets, new arrivals, low priority) that evolves with the signals.
- plateforme Google Ads Performance Max Multi-channel bidding and delivery (Shopping, Search, YouTube, Display) driven by Google's ML.
- outil Google Merchant Center Product feed powering Performance Max; the richness of attributes conditions neighborhood scoring.
How it runs, concretely
For ops teams-
1Structure product attributes Data / e-commerce team
Provide type, brand, price, seasonality at the item level so neighborhood scoring has something to compare.
-
2Score by neighborhood effect SmartScoreAI (tool)
The model estimates a product's potential from its peers, which gives a score to new arrivals with no history.
-
3Re-segment continuously Dynamic Segments (tool)
Products move between categories (high potential, safe bets, new arrivals, low priority) as the signals change.
-
4Steer budget and tROAS at the account level Media team + tool
Spend is reallocated toward the campaigns that perform. The team sets the ROAS thresholds, the tool adjusts.
Product attributes and signals (type, brand, price, seasonality, neighbors' performance). Without rich product attributes, the neighborhood effect does not work and new arrivals stay invisible.
How your customers perceive this type of use
Sourced studiesLe pricing algorithmique est le terrain le plus inflammable : 68% des consommateurs disent se sentir leses quand les marques utilisent le pricing dynamique et 80% jugent plus dignes de confiance les marques aux prix constants (Gartner, 2024). L'equite percue varie selon le secteur : le pricing dynamique n'est juge juste que par 33% a 40% des repondants selon qu'il s'agit de concerts ou de cinemas (YouGov, 17 marches). Le prix personnalise par les donnees individuelles est le plus rejete : 47% des Americains s'y opposent fermement (Consumer Reports, 2024).
Acceptance conditions
- La constance des prix comme signal de confiance : 80% jugent plus fiables les marques aux prix stables (Gartner 2024)
- Le secteur conditionne l'equite percue : le pricing dynamique est mieux tolere pour les cinemas (40% le jugent juste) que pour les concerts (33%) (YouGov 2024)
Red lines
- Le pricing dynamique percu comme abus : 68% se sentent leses (Gartner 2024)
- Le prix individualise a partir des donnees personnelles : 47% d'opposition ferme (Consumer Reports 2024)
- Les frais caches et hausses imprevues, vecus par 79% des consommateurs sur un an et associes a la perte de confiance (Gartner 2024)
Sources: Gartner 2024 · YouGov 2024 · Consumer Reports 2024
How to replicate
Inference, not sourcedData prerequisites
- Rich product attributes at the item level (type, brand, price, seasonality)
- A clean Shopping feed
- Consent Mode v2 in the EU
Org prerequisites
- A catalog with many new arrivals and long-tail items
- A third-party orchestration/scoring tool on top of PMax
Possible stack
- Google Ads Performance Max + a product scoring and orchestration tool (smec or equivalent) + an enriched product feed
The plan, step by step
- Step 1Enrich product attributes at the item level (type, brand, price, seasonality) and clean the Shopping feed.Deliverable: Enriched and validated Merchant Center feed.
- Step 2Plug in the scoring and orchestration tool on top of Performance Max.Deliverable: Neighborhood scoring active across the whole catalog.
- Step 3Set up dynamic segments and a campaign structure by strategic category, with tROAS per segment.Deliverable: Segmented PMax campaigns in delivery.
- Step 4Let the orchestration reallocate budget and tROAS, then read revenue and ROAS on a test category against static bucketing.Deliverable: Revenue/ROAS comparison vs the static approach and a scaling decision.
First step: Identify the new arrivals and long-tail products with no history, then test neighborhood scoring to give them visibility in PMax.
The same pattern, proven in other industries
Allwyn
AI-driven omnichannel programmatic buying with retail data activation
Nivea
genAI testing and optimization of creative elements + automated localization at scale
Clarins
AI-automated sales campaign plus dynamic catalog ads, evaluated on incrementality via Conversion Lift against last-click
Sources
- S1 River Island - smec case study (SmartScoreAI, Dynamic Segments, Performance Max) Interested party archive pending
- S2 smec finalist for the European Search Awards 2026 (Best Use of AI in PPC, with River Island) Interested party archive pending
An error, newer info, a source?
This page lives on its accuracy. If a figure has moved, if the deployment has changed, or if you have a higher-quality source, tell us. Every sourced correction is verified before publication.