In September 2026 I shared a series of LinkedIn posts that distilled research on B2B marketing, synthetic data, decision‑making habits, and AI‑augmented content pipelines.
Only 32% mentioned researching anyone who wasn't already a customer.
In 87% of cases the target buyers weren't the main input.
92% pull insights from sales calls.
The twins showed less spread than the real humans in 154 of 164 outcomes. 94%.
- Decision‑making: 87% of big marketing decisions were gut calls; target buyers were not the main input.
- Decision: Stop relying on informal conference chats; they don’t scale.
- Decision: Invest in systematic non‑customer research to close the gap between claimed (76%) and actual (32%) effort.
- Decision: Avoid synthetic panels for message testing because they mask edge‑case insights.
- Decision: Adopt a three‑layer Claude setup (Projects → Skills → versioned files) to enforce quality and guard‑rails.
- Decision: Publish pricing openly; LLMs now surface it regardless.
- Decision: Differentiate on “determinants” (importance + differentiation) not just qualifiers.
- Decision: Prepare product info for AI purchasing agents.
- Insight sources: 92% pull from sales calls, 85% interview customers, but 53% research the broader market only once a year or never.
- Synthetic panels: twins reduced variance in 154/164 outcomes (94%) and correlated with real answers at only 0.20.
- Checkmarx AI workflow: Projects store context, Skills generate assets, Brand skill acts as a guardrail, not a style guide.
- Token budget: $150 to $200 a month per PMM in tokens.
- CXL AI‑native Marketer cohort starts September 29.