Case Study: Retail

Global retailer expands to new international market with Cerebra data-driven merchandising


Large-scale retail company seeks sales & trend forecasts for merchandising localization for international expansion for customers with different tastes

Cerebra reduced inventory at hand by 35% for the new market allocation over 6 months while driving sales by 23% during COVID with improved assortment

InCerebra successfully predicted trend shifts and rankings for +73% of the SKUs carried, and directed creation of 12 new SKUs based on trend shifts

Cerebra takes the heavy lifting of modeling every customer’s individual demand into predictions, and aggregating and standardizing the results into actionable takeaways


This Cerebra client is an international fashion retailer with e-commerce and brick and mortar stores, with +4000 products offered. It produces and also buys apparel and appeals to youth and business casual segments. The company has strong traction in its core geographies and is seeking to expand into North and South American markets.


The company collects data from a variety of sources ranging from e-commerce reviews, user interviews, return data, and customer behavior. However, given the breadth of 4000+ SKUs and sales patterns in 17 countries, most of the signals that show up-and-coming trends for specific regions or consumer groups go unnoticed.

The merchandising team has a strong grasp over the top 150 performing SKUs, however, the long tail and trend shifts often yield missed opportunities. The trends that are seen during manual inspections on Google Analytics, FB Audience Targeting, and the internal CRM are bubbled up to the merchandising team for planning inventory and discounts, along with developing improved assortments for the upcoming seasons.

The most significant sources of untapped signals are from:

  • aggregating text reviews, from both returns and product reviews of their SKUs and competitors
  • matching product trends to audiences coming from specific channels and regions
  • identifying which products create recurrent buyers and high conversions for which segments
  • collecting user journey behavior on e-commerce to identify why certain products don’t convert

The company has tried traditional text-mining and social listening solutions to improve merchandising, however, the results did not yield direct action courses that can be turned into critical opportunities and did not incorporate projections into their production, replenishment, and sales workflows. In order to drive sales in a competitive marketing expansion moment, the company’s Innovation Team approached Cerebra to synthesize signals from text, user behavior, past sales, and ad conversion. 


Cerebra solves the “Why Them Now?” problem with its proprietary Decision Engine. Cerebra automatically extracts the segments that are most reactive to certain products and given the shifts of the spending trends of each segment, recommends critical merchandising actions driven by sales forecasts. The forecasts span from 2 weeks to 6 months, with different granularities for a variety of applications.

The Cerebra Dashboard is the single source of truth for the merchandising teams to evaluate the performance growth trajectory of every product in real-time. The Cerebra Engine benchmarks all products carried on the e-commerce website against overall product categories, based on urgency and growth delta. Along with current sales performance, each product insight shows behavioral alerts from the customer base that signal future growth opportunities or potential for declining performance.

By receiving customer alerts in real-time for up-and-coming products, the merchandising team was able to identify which products they should focus on for driving customer acquisition and conversion, while improving the assortments and offers to engage the consumer base with the newest trends they are interested in.


Cerebra’s Decision Engine ingests structured or unstructured data in any format: numeric, categorical, time series, text-based, or image-based. The system translates each stream of data into a signal, and groups the signals into customer cohorts and product groups that exhibit similar behavior in relation to each other. The behavioral forecasts for each of these groups are compared against the core KPIs of the company, such as acquiring new customers, reducing losses, improving customer satisfaction, and decreasing stale inventory. Cerebra identifies patterns and causal effects across all disparate signals in real-time, providing recommendations tailored for each role, based on long-term forecasting. The users can identify every factor that impacts the performance of their products, internal and external – along with the reasons why.

The insights adapt to changing customer habits and market landscape and automatically improve based on the results of the actions taken by the company.


Decreased stale inventory allocation. Cerebra reduced the inventory at hand by 35% through reallocating across warehouses in 4 countries based on trend forecasts, while driving sales through better matching the assortments carried for each consumer segment.

Trend-oriented sales growth. During a global pandemic, Cerebra supported 23% growth in sales for the markets it was deployed in. This was accomplished by successful trend forecasts of +73% of the SKUs at hand with precise number of sales, and the best channels to reach the target segments.

Rapid Integration. Cerebra started showing initial results within 4 weeks of deployment, and incrementally improved the recommendations through iterative additions of new data sources, such as ad spending across channels, social listening sources, and website engagement data.

Demand-driven merchandising. The client developed 12 new SKUs based on the product trends Cerebra detected from external data sources, which will be released over the next two seasons.