The default mental model is that Visibility leads Favorability — get yourself named in more AI answers, then defend the way the model talks about you. The Q1 2026 data across the Lantern 200-brand sample inverts that model for two specific categories. In beauty and in fitness and recovery, brands with above-median Favorability and below-median Visibility outperformed brands with the opposite profile on downstream AI-assisted conversion.
The Q1 2026 scatter
Where Favorability beats Visibility
In beauty, buyer queries skew toward ingredient-led and routine-led questions. The model is reasoning about fit — and the words in the descriptor cloud do most of that reasoning. In fitness and recovery, the same logic applies through different vocabulary: descriptors like "physical-therapist-recommended" or "clinical-evidence-backed" sit at the top of the descriptor cloud for top-Favorability brands.
Five anonymised case studies
Case 1 — Beauty, Brand A
Q4 2025 baseline: Visibility 49%, Favorability 62. Offsite work — Trustpilot reviews, dermatologist YouTube placements, descriptor-cloud remediation on three "watch" terms. Three months later: Visibility 51% (essentially flat), Favorability 81 (+19), AI-assisted CVR rose 28%. Visibility did not move; the answer landscape just stopped warning shoppers off.
Case 2 — Fitness, Brand B
Q4 2025 baseline: Visibility 38%, Favorability 71. Lantern flagged a Long-term External Signals gap ahead of Visibility work. The Offsite push moved Favorability to 85 and lifted Brand AI Health by 6.4 points without a single Engine-bundle ship.
Case 3 — Coffee, Brand C
Q4 2025 baseline: Visibility 28%, Favorability 79. High Favorability, low Visibility — the right move was Visibility-led: Catalog Engine bundles across whole-bean PDPs and a comparison-content sprint. Visibility moved to 41% in one quarter. AI-assisted CVR followed Visibility, not Favorability — confirming the category-effect.
Case 4 — Apparel, Brand D
Q4 2025 baseline: Visibility 22%, Favorability 65. Most of Brand D's missing Visibility traced to feed attributes. Five weeks of Catalog work moved Visibility to 34%. Apparel rewards presence first.
Case 5 — Beauty, Brand E (the counter-example)
Q4 2025 baseline: Visibility 58%, Favorability 54. The team pushed Content-agent Engine bundles to lift Visibility further. Brand AI Health gained 1.2 points; AI-assisted CVR did not move. The descriptor cloud was the binding constraint.