Scout
wiki/decisions/2026-05-20-balanced-validation-target-matrix.md
ADR: Balanced Validation Target Matrix
Date: 2026-05-20 Status: Accepted
Decision
Scout validation must use a balanced target matrix rather than overfitting to B2B SaaS examples.
Primary targets:
| Segment | Company | Primary Use |
|---|---|---|
| Private B2B SaaS | Algolia | PRISM, company, careers, blogs, docs |
| Private B2B SaaS | Constructor | PRISM, competitor intel, company, careers, blogs/docs |
| Private retail commerce | L.L.Bean | ecommerce, product catalog, careers, company |
| Private retail commerce | Patagonia | ecommerce, product catalog, company, sustainability/blog content |
| Public company | Adobe | company, investor, careers, blogs/news |
| Public company | Home Depot | public retail, investor, careers, product catalog, news |
| Public / hard-site retail | Estée Lauder | hard-site product/category fallback and blocked-page handling |
| Public travel / airline | British Airways | travel company intel, careers, research, website-quality |
Secondary targets: Nike, Amplience, Salesforce, Intuit.
Implementation
The executable source of truth is scout.core.platform.targets; documentation mirrors that registry in docs/target-matrix.md.
Consequences
- Company/PRISM tests should use all primary targets.
- Product tests should cover L.L.Bean, Patagonia, Home Depot, Estée Lauder, and Nike.
- Investor tests apply primarily to Adobe and Home Depot, with Estée Lauder/British Airways only where public-parent evidence is useful.
- Estée Lauder is a hard-site validation target, not a private retail target.
- British Airways is a travel/research target, not retail commerce.