Visibility metrics are meaningless unless their numerators, denominators, deduplication rules, and sampling windows are documented openly. Here is how Wenjian measures AI brand perception.
Deduplication rule: Multiple occurrences of a brand within a single answer count once, preventing verbose answers from inflating visibility.
Position scope: Computed specifically on comparative and shortlist-style buyer prompts where ordered recommendations exist.
Critical distinction: Being named in prose (Mention) and having your URL cited as evidence (Citation) are tracked as separate metrics.
In cross-engine baseline research conducted by Wenjian Research (anchored to the Q3 2026 90-day observation window), sampling follows explicit parameters:
Because LLM outputs are probabilistic and web indices evolve daily, Wenjian conducts period-over-period comparisons using fixed prompt sets and holdout control prompts rather than isolated single-query snapshots.
| Baseline (T0) | Prior to website or content updates | Lock agreed buyer prompt map and run multi-round sampling across target engines |
| Initial Check (T+7) | ~7 days post-publication | Verify crawler indexing and monitor initial footnote citation pickup |
| Mid-Cycle Check (T+14) | ~14 days post-publication | Re-sample identical prompts to spot responsive engines and remaining gaps |
| Cycle Review (T+30) | ~30 days post-publication | Full period-over-period evaluation of citations, framing accuracy, and next-step priorities |
No fixed-ranking promises: Generative answers rotate retrieval sources dynamically; fixed organic answer slots cannot be bought or promised.
No conflated attribution: When on-site updates and external media run concurrently, website upgrades are evaluated on verifiable process metrics (crawling, indexing, and domain citation pickup).
No unverified technical gimmicks: Rather than chasing trendy files that major search engines confirm have zero ranking or AI Overview impact, we invest engineering effort in visible factual prose and clean HTML.
You do not need a finalized GEO roadmap to begin. Share your brand, priority market, and buyer scenarios—our research and consulting team will review real AI answer observations with you to identify what matters first.