Luke Allen
May 4, 2026
I spent two days at BrightonSEO last week: the world's largest search marketing conference, 5,000+ people, Brighton seafront. The seagulls stole my chips on day two. The market intelligence was worth it.
Here's what stood out.
The Writesonic CEO, Sam Garg, said something in his Day 2 keynote that I keep coming back to: there are now 100+ different AI and web visibility monitoring tools, many of which launched in the last six months alone.
Most of them track. Some of them measure. Almost none of them act.
His framing: "Execution is the moat." Not data. Not dashboards. Not the ability to show a brand where it's invisible in AI answers (every tool does that now). The moat is what you do after the data lands.
When I walked the floor, the competitive breakdown looked something like this:
Three vendors were operating as full-cycle, integrated solutions: Writesonic, Profound, and Uberall . The rest were monitoring, dashboards, reporting.
Only one vendor was focused on Knowledge as a layer at all: Waikay (What AI Knows About You). Their knowledge layer is built from public web crawl, though no proprietary/dark data that lives inside your organization but hasn't made it onto the web.
I didn't see anything at the show approaching GEOforge's Knowledge Base concept.
That's not triumphalism. It's a gap in the market. The industry has converged on tracking what's visible. Nobody is solving for what's invisible: the proprietary expertise, call transcripts, SME knowledge, and internal data that should be shaping how AI engines understand your brand, but isn't being surfaced anywhere.
Jack Lingard from AIP Media made a point that I thought was one of the most practically useful things said across both days.
GA4 only captures directly tracked conversions. But GEO drives value that attribution tooling can't see: return visits, referral lift, direct traffic growth. These don't show up as clean conversion events. They accumulate in the background.
AIP quantifies this using what they call a Revenue Influence Factor (RIF):
RIF = Total Impact ÷ Directly Attributable Revenue
Their example from the keynote:
Tracked Attribution (GA4): £100k
Return users + Referral + Direct Lift (modeled): £80k
Total Impact: £180k
RIF: 1.8×
The methodology uses regression and controlled comparisons to isolate causal effect, not just correlation. The signal inputs are referral traffic uplift, direct traffic growth, branded search volume, and topical coverage expansion. When all four trend together, you can make a credible, defensible case that GEO is the cause.
For anyone still trying to justify GEO investment to a sceptical CFO: GA4 underreports your impact. The actual number is closer to twice what the dashboard shows. And the data exists to prove it.
Another line from Sam Garg's keynote worth sitting with.
For years, SEO was primarily a content production exercise: write more, optimize better, rank higher. AI has dismantled that model. Content production is automatable at scale. The strategic question has shifted from what do we publish to how do we build systems that keep us visible as AI intermediates search.
His architecture for this: one orchestrator agent, then a team of specialist agents (a content agent, a design agent, an SEO agent), each with access to specific tools, reporting back to a single goal. You set the objective ("increase AI visibility by 15% this quarter") and every task traces back to it.
Profound's keynote took this further, introducing the concept of the Marketing Engineer: a new job title for someone who builds AI systems and agents instead of running campaigns manually. The premise is that AI now handles the execution layer of marketing, and someone needs to own the infrastructure that makes it work. They argued SEOs are the natural fit because they already think in systems, work with data, and have the technical instincts.
Whether that job title takes hold or not, the direction is right. The most valuable marketing hire over the next few years won't be someone who writes faster. It'll be someone who builds better.
Pablo Lopez from DEPT walked through a methodical process for generating the actual prompts buyers are submitting to LLMs, working backwards from keyword research.
The workflow:
The output is prompts like "I've struggled with wrist pain for months while working in the office, what keyboard would you recommend?" rather than a keyword like "best ergonomic keyboard." That's the shift GEO demands. LLMs respond to natural language queries with context. Keyword-matched content is answering a different question than what buyers are actually asking.
For completeness, here's a snapshot of who I spoke to and what they're building:
Writesonic: Broad LLM platform coverage (10+ AI tools), traditional SEO tooling. Public web data only. No proprietary knowledge base concept.
Profound: AI search visibility and analytics. Well-funded ($155M raised, $1B valuation). Locality GEO focus.
Linkuma: GEO platform covering LLM Citations, AI Overviews, Brand Mentions, Entity Authority.
Waikay: Knowledge graph / topical knowledge focus. Public crawl only, no proprietary data layer.
Rank Math: AI visibility + WordPress-only publishing.
Promptwatch: AI visibility and prompt improvements. No content or publishing capability.
Uberall: Locality-focused GEO.
RankingSuperior: Topical authority + citations for SEO agencies and brands.
Yoast: Structured data and AI-readability optimization.
The space has matured faster than I expected. Visibility monitoring is table stakes now. Citation building is becoming standard. The proprietary knowledge layer — the one that separates what AI engines know about your brand from what they could know — remains unsolved by everyone I saw at the show.
The CEO of Writesonic said execution is the moat.
After two days on the floor, I'd say he's right. I just didn't see anyone who's built the full thing.
My role is to lead commercial and GEO strategy - helping brands get found in the age of AI search via mentions on ChatGPT, Gemini, and the AI platforms. My mission in life is to then trace it out to real revenue outcomes.
