What challenges did Lush face before AI implementation?

Products without barcodes or labels slowed checkout because staff had to identify items manually. Manual identification created bottlenecks during peak periods and made onboarding harder for new employees. Manual entry also affected billing accuracy and inventory management.

What was the impact of the AI solution?

Queue times at the busiest times of year fell to around 3 minutes. Since 2023 the deployment reduced packaging and demonstration materials, including 220,000 ballistic bags, 440,000 liters of water, and products worth £352,000. The system improved billing accuracy, inventory management, and onboarding support for new employees. The approach also supports sustainable and inclusive retail operations.

Source

GCP

Lush Lens retail image recognition at checkout using Vertex AI

Use case typeComputer vision checkoutUpdated Jan 1, 2024

Lush built Lush Lens, an AI-powered image recognition system, to identify packaging-free products at checkout and reduce manual product entry. The system was integrated with mobile point-of-sale devices so staff can point a camera at products and automatically add items to a customer basket. Google Cloud Vertex AI supports model development, training, and deployment, while Cloud Storage hosts the image library used for training.

Organization: Lush
Industry: Retail
Location: United Kingdom
Published: January 2024

Reported outcomes

+49%

packaging-free productsOther quantified impact

3 minutesqueue times at busiest times440,000 literswater for product demonstrations352,000 GBPproducts worth reduced220,000 countballistic bags reduced

Strategic outcomes

Cost efficiencyStreamlined checkout for barcode-free productsEmployee experienceShortened onboarding for new employeesCustomer experience & trustImproved checkout experience during peak periodsSustainability & ESGReduced packaging and demo material useOther strategic outcomeSupported employees with learning differences

Primary read

Use case focus

Showing 2 of 2

1Computer vision checkout
2Inventory optimization

Products without barcodes or labels slowed checkout because staff had to identify items manually.
Manual identification created bottlenecks during peak periods and made onboarding harder for new employees.
Manual entry also affected billing accuracy and inventory management.

Lush captured hundreds of images of each product from multiple angles to build training datasets.
The company trained machine learning models on Vertex AI to identify unpackaged products regardless of shape, color, or orientation.
The trained models were integrated with mobile point-of-sale devices and the Lush Lens checkout workflow.

Technologies

Vertex AI Cloud Storage

Queue times at the busiest times of year fell to around 3 minutes.
Since 2023 the deployment reduced packaging and demonstration materials, including 220,000 ballistic bags, 440,000 liters of water, and products worth £352,000.
The system improved billing accuracy, inventory management, and onboarding support for new employees.
The approach also supports sustainable and inclusive retail operations.

Architecture

Lush Lens uses image recognition at checkout. Images of products are stored in Cloud Storage and used to train and deploy machine learning models on Vertex AI. The models are integrated into mobile point-of-sale devices so staff can scan products with the till camera and automatically add items to the basket.

Sources & evidence1

Groundedness: 5/5Type: Customer StoryPublished: Jan 1, 2024Publisher: Google CloudEvidence: PrimaryConfidence: High

Primary source

AI-generated summary. Verify important details with the linked sources before relying on this case.

Explore related AI use cases

Customer evidence