quality_assurance · ecommerce · workflow
Procter & Gamble adopts Labelbox as enterprise-wide AI training data platform
P&G collects large amounts of unstructured data (images, videos, and text) across prototyping, testing, and manufacturing, and needed a central platform to label this data for AI model training while maintaining strict data privacy, security, and full process transparency.
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 · Collect unstructured data
P&G collects large amounts of unstructured data in the form of images, videos, and text throughout their prototyping, testing, and manufacturing process.
Tools used
Labelbox
Outcome
After three years of expansion, Labelbox became an enterprise-wide corporate solution deployed across all P&G business units globally, described as a transformation in how they work and driving faster model iteration and time to market.
What failed first
Previous labeling partners operated as black boxes — data was sent out and returned labeled with no visibility into the labeling process or how data was being enriched.
Results
Time savedthree years
Grounding & classification
Source type: vendor customer story
20 fields verified against source quotes.
computer visionquality inspectionfailure mode describedhuman review describednamed customerproduction runtime claimedtools describedworkflow describedmanufacturingcycle time reductionemployee productivityvendor customer storydata entry opsquality assurancedocument to recordhuman review queue