quality_assurance · saas · workflow
How Criteo builds better contextual advertising products with AI
Criteo's Publisher Content Analysis team could not reliably improve label quality or efficiency for their ML models, and managed unstructured image data through Excel spreadsheets with label quality defined through many back-and-forth internal emails.
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 · Scale ML annotation need
Criteo needed to quickly scale their ability to improve AI training data for ML models across multiple project teams.
Tools used
Labelbox Annotate
Outcome
Criteo immediately saw a 40% gain in annotation delivery speed with comparable increases in annotation quality, and achieved a massive reduction in daily back-and-forth between product and ML teams.
Results
Volume40%
Grounding & classification
Source type: vendor customer story
21 fields verified against source quotes.
computer visiondocument classificationproduct cataloghuman review describedmetric backednamed customersource backedtools describedworkflow describedmediasoftwareaccuracy improvementemployee productivitythroughput increasevendor customer storymarketing opsquality assurancehuman review queue