Why audit now
The research backdrop is consistent across sources. Around 73 percent of the buying journey happens anonymously before vendor contact. Generative AI is now a top-five discovery channel per McKinsey's Global B2B Pulse, with usage in vendor research at roughly nine in ten buyers. And analysis of the AI-mediated buying process notes traditional attribution captures barely a quarter of the journey. You cannot manage a funnel you have never looked at from the buyer's side.
The four-layer audit
- Layer 1: The machine longlistAsk ChatGPT, Claude, Perplexity, and Gemini the questions your buyer would ask: best [category] firm for [ICP] in [market], alternatives to [known rival], how to choose a [category] partner. Record every named firm and every cited domain. Run each prompt twice; variance is itself a finding. Full context on why this layer matters in our AI Search briefing.
- Layer 2: The search surfaceCheck the classic queries plus AI Overviews. TrustRadius found 72 percent of buyers encountered AI Overviews during research and 90 percent clicked at least one cited source. Note who owns the answer box for your category questions, and whether the cited pages are yours, a rival's, or an aggregator's.
- Layer 3: The human proof layerReviews, directories, communities, and podcasts where your category is discussed. 84 percent of buyers use peer recommendations and reviews early. Search your firm and your top two rivals in each. Count mentions, recency, and sentiment. Thin here usually explains stalled late-stage deals.
- Layer 4: The founder recordAudit the founder profile as a stranger: last 90 days of output, coherence of themes, whether a skeptical CFO could build conviction in five minutes. Individual profiles out-cite company pages in AI results, so this layer feeds Layer 1 directly.
Scoring the gap
MagnetizeX builds founder visibility systems for B2B firms.
Score each layer 0 to 5 for yourself and your two closest competitors: 0 means absent, 3 means present but generic, 5 means the reference everyone else is compared against. Twenty points available. In our client work the pattern is stable enough to state as a rule of thumb: firms losing deals they should win usually score under 8 while their rival scores mid-teens, and the deficit concentrates in Layers 1 and 4, the two layers that compound.
The audit rarely reveals a marketing problem. It reveals an absence problem. You cannot be chosen from a list you are not on.
What to do with the map
Fix in reverse order of difficulty. Layer 4 is fully in your control and feeds everything else. Layer 3 responds to a deliberate review-and-community program within a quarter. Layers 1 and 2 move slowest because they aggregate the others; they are the scoreboard, not the lever. If the audit shows a positioning problem underneath the visibility problem, and it often does, start with how models summarize your firm before producing more content. Volume on top of fuzzy positioning just makes the fuzz louder.
We will re-run this framework quarterly against public data and publish what shifts. The media landscape it depends on is moving fast, as covered in PR without press, so treat any single audit as a snapshot with a short shelf life.