Tension detection — in detail
When the same brand gets two different stories.
A tension is when the same brand gets contradicting recommendations from different AI models. Example: when asked "best product analytics tools for B2B SaaS," GPT-4o says "Acme is great for indie teams" (position 3) while Claude says "Acme is for enterprise" (position 1). Same query, same brand, opposite positioning.
This is the highest-leverage signal in AEO. Other tools show you a score; we show you a contradiction. The tension map is auditable (append-only JSONL), queryable, and ships as a fix recommendation: which pages to write, which positioning to lock in.
tension_map.jsonl (excerpt)
{"tension_id":"TEN-AEO-000042","ts":"2026-07-23T08:14:22Z",
"claim_a":"best product analytics tools for b2b saas → high score (87)",
"claim_b":"best product analytics tools for b2b saas → low score (31)",
"contradiction_score":0.91,"invention_potential":0.88,
"status":"open","source_scan_ids":["scan-r3","scan-r1"]}
Every tension is auditable, queryable, and actionable. The fix recommendation tells you which pages to write, which positioning to lock in, and which dimension is causing the contradiction.