Real-Time Discovery for a Dynamic Toy Universe

Toys"R"Us is a unique retail environment - toys are driven by micro-signals like character affinity, age nuance, gifting context, play patterns, and emotional resonance.

PSYKHE AI translated that nuance into real-time, per-shopper product ranking, ensuring the right products surface to the right customers in the moment, across a broad and diverse assortment.

Gary Haas - Vice President, E-Commerce

About the Company

Toys"R"Us, part of WHP Global, is one of the world’s most iconic toy retailers, offering a vast selection spanning educational toys, collectibles, games, dolls, vehicles, and licensed character products.

Unlike traditional retail categories, toy discovery is deeply contextual: the right product depends on the child’s age, developmental stage, character interests, play style, and the context of the purchase - whether it’s a birthday gift, a toy for outdoors, a holiday surprise, or everyday play. Helping shoppers navigate that complexity and surface the perfect toy quickly is central to the Toys"R"Us experience.

The Challenge

Toy discovery is a unique journey. A single shopper might be searching for:

a birthday gift for a 6-year-old girl
a character-driven toy for a boy
a STEM learning toy

‍
Legacy merchandising approaches struggle to capture this nuance at scale.

The core challenge was:
How do you surface the most relevant toys for each shopper instantly - across thousands of SKUs, characters, ages, and play styles - especially when a shopper could be shopping for multiple children?

‍

Our Approach

Discovery in specialty retail such as toys requires balancing two competing forces: precision and exploration. Shoppers may start with a rough idea - a character, an age range, a type of play - but toy sessions tend to expand quickly. Parents and gift buyers often browse across categories and end up purchasing for multiple children or occasions in the same visit.

PSYKHE AI built specialist toy embeddings that model signals like character affinity, developmental stage, and play pattern - the kinds of micro-signals that actually drive toy choice.Those embeddings feed a real-time, per-user reinforcement learning system that ranks products per-shopper, continuously adapting as intent reveals itself.

As shoppers browse, hover, click, and move between categories, the system recalculates ranking in real time - responding to micro-signals within the session instead of forcing a single interpretation of intent too early.

The result is a discovery experience that stays both precise and expansive: surfacing toys that match the shopper’s immediate context while preserving enough range for exploration across Toys"R"Us’s broad assortment.This rapid reinforcement learning is especially important in gifting scenarios, where shoppers may be buying for several children with very different interests. Instead of narrowing the catalog too quickly, the system adapts as the session unfolds - helping the right toys surface at the right moment.

What's Next

Expansion of real-time personalization across additional site surfaces, including Search and PDP carousels.

explore next case study

Contact us to discuss your Enterprise’s specific needs and get an in‑depth understanding of how psych-based AI models can deliver the experience of a personal shopper at scale.

Get in touch

Thank you! We will get back to you ASAP!

Oops! Something went wrong while submitting the form.