A data provider organizes people or processes to produce labeled or curated data for a customer. In preference-data work, providers are scarce suppliers, so procurement depends on credibility, budget, timing, and operational capacity. section 11.2.5
Working with a provider requires managing several practical risks:
- Source vendors or internal annotators, recognizing that suppliers may prioritize large budgets, influential brands, and future revenue. section 11.2.5
- Settle contracts carefully: delivery failures, misleading customer claims, administrative snags, and restrictions on open-sourcing acquired artifacts can materially affect the project. section 11.2.5
- Translate the purchased task into detailed instructions covering goals, corner cases, and priorities before labeling begins. section 11.2.5
Multibatch preference-data contract
A ramp-up period narrows goals and methodology before larger batches proceed. Early batches may contain more quality failures and be discarded; one illustrated timeline concerns a contract of roughly $500,000.
