How Walmart uses Labelbox data to improve their natural language models
Walmart's data science team needed faster ways to annotate conversational text from shopping chatbots and label inventory images, but their existing tech-enabled BPO vendors operated as a black box with poor transparency, sparse visibility into data quality, and no tooling for stakeholder collaboration.
How it works
Common implementation structure
How this type of workflow is generally built, generalized across documented cases — not tied to any one vendor's stack. Click any stage to read what happens there. Specific products that implement these stages appear in “Tools commonly seen” below.
Stage 1 · Label data need identified
The data science team wanted faster ways to annotate conversational text from shopping chatbots and label inventory images for object detection and classification models.
Tools used
LabelboxLabelbox AnnotateLabelbox Python SDKNER
Outcome
Walmart saw labeled data accuracy improve by an estimated 25%, achieved 95% accuracy in labeled data via Labelbox's Labeling Services, and reduced turnaround time by 25% compared to similar services, with full visibility into the labeling pipeline.
What failed first
Walmart previously relied on tech-enabled BPOs for labeled training data, but the arrangement proved suboptimal: BPOs lacked transparency, omitted key analytics, and prevented in-house subject matter experts from closely collaborating with external service providers.