Articles from inside the work.
Shorter articles, field notes, and audit teardowns from real client work.
Latest
Every piece here is written from inside the work. The patterns come from real decisions, live tools, and the occasional awkward discovery.
- Tools review30 August 202616 min
Best SEO tools review 2026: the free and paid tools we actually use in the age of AI search
Our 2026 review of the SEO tools we use: crawlers, Google's own AI reports, Ahrefs, AI citation trackers, crawler controls, and the AI assistant layer. What is free, what to pay for, and what to skip.
Answer engine optimisation for B2BFIG.0 / THE 2026 STACK - Article26 August 2026
Publishers have more AI visibility options than they think
Training and retrieval crawlers are now separate, and Google offers passage-level snippet control. Our research found publishers split sharply between fully open and fully closed, with almost none occupying the middle ground.
Answer engine optimisation for B2BFIG.0 / PLUGIN 
- Myth-check18 June 202611 min
Why llms.txt does almost nothing for AI search
llms.txt has a narrow role in agent readiness. It is not a shortcut to AI Overviews, ChatGPT citations, or Perplexity visibility.
Answer engine optimisation for B2BFIG.0 / THE FETCH LOG - Field guide30 May 20269 min
How not to get burned wiring AI into your marketing stack
Connecting an AI agent to your tools widens what can go wrong. Scope tight, default to read, keep a human on writes, and log everything.
AI-native marketing operationsFIG.0 / BLAST-RADIUS CHECK - Field guide30 May 20268 min
The connectors worth wiring first
You don't need 15 MCP connectors. You need the few you'd query weekly, that are safe to connect, and that earn their place in the model's context.
AI-native marketing operationsFIG.0 / WIRE-FIRST TEST - Field guide30 May 20269 min
What MCP actually changes for a marketing team (and what it doesn't)
MCP lets an AI assistant read across your marketing stack. The practical win is faster answers, not an AI running your campaigns.
AI-native marketing operationsFIG.0 / READ-WRITE LINE - Evidence review29 May 20268 min
The passage-first content model: what's real, what's ritual
Passage-first is real. The magic 40 to 60 word answer block is folklore. Here is what the evidence actually supports.
Answer engine optimisation for B2BFIG.0 / THE GRADER
Looking for something specific?
If you can't find what you're looking for, email robin@weareallconnected.co.uk and we'll point you to it, or tell you we haven't written about it yet.