Caveats — only three causal findings; everything else is correlational or vendor-claimed

Honest gaps:

1. Only three findings rise to genuinely causal:

   • Reviews → revenue (Luca regression-discontinuity Yelp study — see sister brief Research brief: SMB widget market difficulty — six ranked factors (June 2026))
   • Page speed → conversion (Google/SOASTA + web.dev A/B partners — Google/SOASTA 2017 — P(bounce) rises 32% at 1s→3s, 123% at 1s→10s)
   • GBP mechanics → local-pack presence (Sterling Sky/Whitespark controlled tests — Sterling Sky controlled tests — GBP posts, geotagged photos, description keywords DON'T move ranking)

   Local-SEO weightings, content timelines, review thresholds, schema CTR lifts, and tool-linkability figures are correlational, survey-based, or vendor self-reports.

2. Survivorship and selection bias are pervasive, especially for content and interactive tools. Treat all agency case studies as a biased sample of winners.

3. Recency and drift. AI Overview citation behavior is changing rapidly; CWV pass rates and review-threshold expectations shift annually. Treat any single-year figure as a snapshot and design tiers to be robust to these shifts.

4. Vertical variation. Ranking-factor weights and review dynamics vary by industry (proximity matters more for home services; reviews + content relevance more for restaurants and professional services). Treat the tiers as universal but acknowledge that within-tier priorities shift by vertical — see sister brief Research brief: SMB widget vertical difficulty — two-axis tiering by industry (June 2026).