Case Study:Improve new product & product assortment merchandising


Improve new product & product assortment merchandising

Anticipating a market trend can make all the difference for retailers . Knowing which new products to bring to market, or how to shift existing assortments historically requires countless hours of merchandisers pouring over blogs, magazines, competitor websites, and more. As new and alternative resources like social media grow in customer relevance, new challenges and opportunities have emerged with identifying new trends.

Client engaged Travix Team  to help design a data & AI powered solution that could improve their existing merchandising, thus maximizing revenues and minimizing unsold inventories.

Client Challenges

Unclear Data strategy
Missing Critical Data
Insufficient AI/ML Talent

Tavix AI Engagement

Data Strategy Coaching
External data sourcing
AI/ML Coaching

Tavix AI Solution

Working closely with the client’s senior leaders, Travix helped develop a strategy most likely to realize optimal results. While a fully automated merchandising process appealed to some, to achieve the highest quality results, and to ensure internal adoption, the solution had to focus on augmenting human merchandisers, not replacing them.

With this in mind, Tavix AI team designed an approach playing to the relative strengths of algorithmic solutions and human merchandisers. First, Travix AI’s web crawlers and scrapers aggregated raw data in quantities impossible for human employees to match. Next, multiple machine learning and AI techniques processed raw content into multiple hierarchical topics, and ranked each by size and rate of growth.

For ease of use by the end user, this output data then integrated with Tableau, where human merchandisers could explore and apply their intuition, creativity, and deep expertise to draw conclusions impossible for modern AI to match in quality.


Due to Tavix ’s explicit “human-AI collaboration” design guidance, senior leaders were able to achieve their goals, while merchandiser staff were able to focus more of their work on value add areas.

3000 daily social media posts from top pages analyzed
20,000 popular web pages regularly analyzed
output hierarchical levels of topic data