BrightonSEO 2026 Just Validated the Three-Layer Approach.

L3 won the room, but it builds on L1 and L2

Jean Dejsuvan's session drew the clearest line we've seen between SEO and GEO: SEO gets you found, GEO gets you recommended. Her three conditions for being recommended - be a brand worth recommending, be easy to recognize and understand, and leave consistent, verifiable evidence across the entire web, not just your own site - is essentially L3 described without the acronym.

StrategiQ's broader takeaway sharpens this further: if your website disappeared tomorrow, could AI still describe your brand accurately? Signals come from PR, reviews, forums, and sentiment - not just on-page structure.

Information gain: the "truffles vs. slop" problem

Several sessions, synthesized by StrategiQ, converged on the same point: generic content is becoming invisible to models that reward novelty and verifiability. Paris Childress's measured lift in AI citations from proprietary insight ("truffles") over derivative content ("slop") is the clearest data point yet that ACU-structured, fact-dense content isn't optional - it's the entry fee for citability.

Accessibility is now a GEO signal, not just an ethics checkbox

Tania Gerard and Damien Robert's sessions made a point we've been arguing for a while: many LLMs read pages through the Accessibility Tree - the same structure screen readers use. Missing alt text, text baked into images, heavy client-side rendering, and low contrast don't just fail disabled users; they fail machine extraction.

L1 gets a name: empathy

Katie Mishner's framing — "empathy is the missing E in E-E-A-T" - and Chloe Blades's "the algorithm gets you seen, psychology gets you remembered" are pure L1, articulated better than we've managed internally. Blades's formula is worth stealing outright: information + emotion = memory.

Ross Simmonds's keynote pushed this into L1+L3 territory directly: create experiences so strong that your audience effectively installs your brand into the model's memory themselves - through what they say about you, not just what you say about yourself.

L2 hasn't disappeared - it's table stakes

Dan Brooks's facet-vs-filter framework (optimize facets with real search demand, don't index filters without it) is classic L2, and the most bookmarked content of the event - a sign that the room still cares deeply about getting the technical foundation right.

What this means for your diagnostic

BrightonSEO 2026 is, in effect, external validation of the layer-scoring logic in 3LA. The practical next step for anyone running audits: look specifically for "dark opportunities" - high L2, low L3 - and pair technical fixes with the off-site entity and brand-consistency work the conference kept returning to. Basic audits are table stakes now; the value is in prioritization and original insight, which is exactly where diagnostic frameworks like this one are supposed to earn their keep.