The Maze: AI visibility is starting to look less like an auction and more like a memory test. A WARC/Charlie Oscar visual shared on LinkedIn puts long-term brand equity at 63% of modeled LLM visibility, versus 22% for marketing spend, 11% for citation volume, and 4% for influencer and PR reach. That does not kill performance marketing. It just demotes it from steering wheel to accelerator.
Brand equity is the big block, not a soft halo. The visible model assigns 63% of AI visibility to long-term brand equity, making it larger than marketing spend, citation volume, and influencer/PR reach combined. That matters because generative engines do not browse like humans scanning ads. They retrieve, summarize, and rank signals from a messy public memory. If a brand is consistently described, trusted, and associated with a category, it gives the machine something clean to grab.
Paid spend still matters, but it does not buy the answer. Marketing spend gets 22% in the WARC/Charlie Oscar model. That is material, but it is roughly one-third of the brand-equity contribution. The commercial lesson is uncomfortable: you can rent attention and still fail to become the default recommendation. In AI search, the expensive part may be building the durable mental file that lets the brand survive retrieval, ranking, and synthesis.
Citation volume is a lever, not the whole discipline. The visible 11% citation slice is where many generative-engine optimization debates are clustering. The original GEO research showed that tactics such as adding citations, statistics, and authoritative language can improve visibility in generated responses, with outcomes varying by domain. Useful, yes. Magic wand, no. Citations help the machine trust and reuse information. They do not replace the harder work of being known for something.
The measurement problem is still messy. The source post does not show sample, market scope, model design, or whether `AI visibility` means mentions, citations, rankings, recommendations, or answer share. That caveat matters. Newer GEO research separates citation selection from citation absorption and finds that different AI search platforms behave differently. So the right reading is not `brand equity equals 63% everywhere`. It is sharper: AI visibility is becoming a brand-system outcome, not a channel-only outcome.
The operator playbook gets less glamorous. Distinct positioning. Consistent naming. Original points of view. Evidence-rich pages. Third-party authority. Clear category language. These are old muscles with a new machine audience. The comments under the post made the same point in plain English: AI does not reward content because it is from a brand. It rewards sources that look authoritative, structured, and worth repeating.
Why it matters: Ecommerce brands are about to relearn a boring truth through an expensive interface. If AI assistants become shopping guides, comparison engines, and category explainers, the winner is not only the brand with the loudest campaign. It is the brand that the web already understands. Performance media can spike demand. Brand memory may decide who gets named when the machine answers.
Sources: LinkedIn source post | GEO paper | Citation absorption paper
