date: 2026-05-15
insights_added: 4
operators_added: 0
patterns_updated: 3
What landed today, 2026-05-15
Four new cards, zero new operators, three patterns extended. Ray and Solis published within 24 hours of each other and arrived at the same structural conclusion about AEO from opposite directions. Schwartz, writing a week earlier, named the measurement problem that makes their convergence land harder. Gupta gave one sentence that changes how you scope a skills library.
Theme 1, the citation layer is external, and one penalty hits everything
Lily Ray (A Google penalty removes a page from every AI surface at once, not just Google search) was working from enforcement patterns. AI systems retrieve content via RAG from Google's index. A Google enforcement action does not stop at Google. It removes the page from the source pool that ChatGPT, Perplexity, and Copilot all draw from simultaneously. Tactics already flagged: self-promotional listicles (49% visibility loss after January 2026 enforcement), comparison page farms, artificial timestamp refreshes. Ray's prescription:
That's getting recommended by everybody else without recommending yourself. And to me, that's where you want to go.
Aleyda Solis (AI search visibility is an off-site corroboration problem with an on-site quality floor, not the inverse) came from the other direction, analyzing citation patterns across five ecommerce subverticals. Her reframe:
AI search visibility for ecommerce is structurally an off-site corroboration problem with an on-site quality floor, not the inverse.
AI cites the page that resolves buyer uncertainty. That is often a third-party guide, a return policy breakdown, a forum thread, or an analyst comparison. Not the product page. The diagnostic job before any on-site sprint: map which off-site corroboration surfaces exist and which are thin or absent. Both cards extend The AEO triangle, presence, relevance, manual-action propagation with a practical sequencing rule. Audit off-site coverage first, check the SEO floor second, then commit to on-site optimization.
- Lily Ray, A Google penalty removes a page from every AI surface at once, not just Google search. A Google penalty removes a page from the RAG pool every major AI surface draws from, turning one enforcement action into a multi-surface traffic loss.
- Aleyda Solis, AI search visibility is an off-site corroboration problem with an on-site quality floor, not the inverse. AEO visibility is structurally an off-site problem with an on-site quality floor. The real optimization work is in third-party environments.
Theme 2, the constructs you measure are dissolving, and the workflows you repeat belong in skills
Eli Schwartz (Keyword position reporting is measuring a construct that is dissolving as Google personalizes results per user) published May 7. His claim: Google has crossed into genuine context-aware personalization using two decades of behavioral data. Same query, same neighborhood, same moment. Two users see different results. Schwartz:
There is no longer a single stable answer to rank for.
Keyword-position reporting is measuring a construct that is quietly dissolving. A dashboard showing "position 4 for [query]" is a statistical artifact of a fragmenting distribution. Personalization currently lives inside Gemini and will expand to AI Mode, then AI Overviews, then all search surfaces. The reporting methodology will need a rebuild before clients feel the damage. This extends Absolute counts + correlated short signals, not stage rates and long loops: position is no longer a stable signal.
Aakash Gupta (If you would paste the same instructions twice, that workflow belongs in a skill, not session memory) is working a different but adjacent problem. After 75 tests of Claude Skills, he landed on one diagnostic:
The test: if you'd paste the same instructions twice, it belongs in a skill.
The primary failure mode across all 75 tests was the description field, which is the routing layer. A vague description means the skill does not trigger in contexts where it would clearly help. This extends Context, not capability, is the bottleneck: recurring instructions should live in skills with precise descriptions, not in session paste buffers.
- Eli Schwartz, Keyword position reporting is measuring a construct that is dissolving as Google personalizes results per user. Keyword position reporting is measuring a dissolving construct as Google expands context-aware personalization across all search surfaces.
- Aakash Gupta, If you would paste the same instructions twice, that workflow belongs in a skill, not session memory. If you would paste the same instructions twice, that workflow belongs in a skill. The description field is the routing layer.