Target’s Revolutionary AI-Based Accessory Recommendation System: GRAM

Target has made a significant leap in enhancing the shopping experience by deploying a generative AI-based accessory recommendation system called GRAM (GenAI-based Related Accessory Model). Designed specifically for the Home category, this innovative system addresses the complexities of pairing complementary products within Target’s extensive retail catalog.

The Challenge of Accessory Recommendations

With a vast array of products, Target faces unique challenges when it comes to recommending related accessories. Shoppers often consider various attributes like color, material, brand, and even intended audience. Traditional rule-based systems, which rely on static guidelines, can struggle with this complexity and lack the flexibility needed to adapt to diverse customer preferences. As a result, the need for a more advanced solution became evident.

How GRAM Works

At its core, GRAM leverages large language models (LLMs) to analyze structured product data. This enables the system to determine which attributes matter most for each core-accessory pairing. By evaluating these attributes and assigning importance weights, GRAM generates relevance scores that shape the recommendations presented to shoppers. For instance, when matching a throw pillow with a sofa, the system focuses on attributes like color and material. In contrast, when suggesting a battery for a toy, it prioritizes compatibility and safety features for kids.

Blending AI with Human Expertise

One of the key features of GRAM is its combination of AI capabilities with a human-in-the-loop process. This approach allows Target’s merchandising team to curate lists of co-purchased or seasonal items. By marrying the scalability of AI with human insights, the system not only improves discovery but also ensures that recommendations align with merchandising goals, enhancing overall customer satisfaction.

Insights from Target’s Data Science Team

Adnan Awow, Senior Director of Data Science at Target, highlights the advantages of GRAM in a recent LinkedIn post, stating, “With GRAM, our GenAI-powered Related Accessory Model, we automatically prioritize hundreds of product attributes such as color, material, and brand to surface the most relevant add-ons first.” This system factors in aesthetic cohesion, enabling the suggestion of visually harmonious combinations—like pillowcases that beautifully complement sheets.

Early Success and Measurable Outcomes

Target has already seen impressive results from GRAM through early A/B testing. Add-to-cart interactions for Home category accessories rose by approximately 11%, and the display-to-conversion rate increased by 12%. Additionally, attributable demand experienced a growth of 9%. Following these promising outcomes, Target fully deployed the system into production, underscoring its commitment to integrating AI into the digital shopping experience for enhanced personalization.

A Scalable Future of Recommendations

The implementation of GRAM marks a pivotal moment for Target in offering scalable and relevant accessory recommendations. By combining advanced AI capabilities with human judgment, the company is not only boosting user engagement but also creating a more seamless shopping experience. The full deployment of this model in April 2025 illustrates Target’s dedication to continually evolving and improving personalization and recommendation quality across its digital channels.

As the retail landscape continues to evolve, solutions like GRAM are set to redefine how shoppers discover and purchase complementary products, promoting a more engaging and efficient shopping journey.

Inspired by: Source