Chris Green: Patents, Bots & the Cannibalization Reversal

Chris Green of Torque Partnership has spent fifteen years watching SEO theory get built on Google's patent literature. In this episode he takes the two load-bearing ones, PageRank and Reasonable Surfer, and asks what survives a web where most of the traffic arriving on your pages was never a person.

Chris started in technical and content, moved through running and managing teams, then into data and analytics. Today he runs global programmes for SEO, GEO, measurement and AI search for Fortune 500 brands, a job he describes as trying to wrangle this stuff and make it understandable and accessible to as much of the org as possible. As he puts it, that is a very different skill set to being an agency SEO. He is based in the UK and publishes his working notes in public on Substack.

In this episode

• PageRank as a principle, not a blueprint. Simulating equity flow is a good way to visualise a linking change. Hedging a rollout decision on that simulation is the risk, because nobody outside Google can tell you what the mechanism now intersects with.
• Reasonable Surfer when the surfer is a bot. Agents arrive directly, never pogo back, and collapse dwell time, click patterns and frustration signals. Chris has watched the analytics and says the behaviour is frankly totally different.
• Bot traffic pollutes the data that pays your salary. Polluted analytics ruins decision making, which eventually ruins the budget and the remit. The hardest case is an agent riding the user's own user agent, session and device.
• Markdown companion files: trivial on a simple site, a nightmare on a monolith. On a clean WordPress install it is almost a non-task. On a million-page stack held together with duct tape, the cost is consistency and ownership across markdown, HTML and feeds, and nothing on the internet will tell you when they drift apart.
• Which agentic standard to back. No protocol has won. The current strategy is to do as much of it as you can and hope you backed the right horse.
• The RAG layer is where the gaming is. Training data moves slowly and is hard to value. Retrieval moves fast, and old link building techniques can still be far more effective there.
• The cannibalization reversal. An isolated client test found three to four selectable links got cited more often than fewer than three, or more than four or five. The operational instruction is keep one and the other, not move one.
• Access logs as the measurement layer nobody is selling. Segment your content, split-test bot hits, classify each hit as training, search or user-triggered, and check the IP range before you trust the user agent.

Chapters

• (00:00) Fifteen years, and the seats Chris has sat in
• (01:18) Do Google's old patents still describe live systems?
• (04:50) Test it and stay skeptical, the SEO's actual core skill
• (05:10) Reasonable Surfer when most of the web is bots
• (07:19) Polluted analytics, ruined decisions, ruined budgets
• (08:17) Markdown companion files and Google's position on them
• (11:01) The trust problem, and the maintenance overhead on enterprise stacks
• (14:28) Bots with wallets, and which standard to hedge
• (17:29) Link placement, link building, and whether the watershed has come
• (20:13) The RAG layer is more ripe for gaming than training data
• (22:57) The tactical win that has always worked: be explicit
• (25:07) The cannibalization reversal, and the three to four link sweet spot
• (26:46) Keep one and the other, not move one
• (29:19) When GEO and SEO pull against each other under separate governance
• (32:58) Proving value: brand mentions as the soft floor
• (33:56) Access logs as the treasure trove
• (34:47) Convergence, and where the SEO still has a claim
• (37:24) Using AI on log files without hallucinating trends
• (41:28) Spoofed user agents, and checking the IP range
• (42:21) Training bot, search bot, user-triggered bot
• (43:11) Torque Partnership, and where to find Chris's work

Notable quotes

"It's kind of like the Wild West in a very small niche and overly techy kind of way."— Chris Green (14:08)

"If you had three to four links that were selectable, you were much more likely to get included in the citation than if you had under three."— Chris Green (25:36)

"One of the other treasure troves of information is access logs."— Chris Green (33:56)

Resources & links

• Torque Partnership, Chris's consultancy: tech, CDN and global SEO programmes
• chris-green.net · LinkedIn · Substack · Bluesky
• Chris on what percentage of your traffic is actually bots
• Chris on why user agents matter, the detail behind the access-log method in this episode
• Chris on rank tracking in LLMs and AI chatbots
• Chris on the agent-facing web in 2026
• Dan Kurtz on bots as a customer segment, the previous episode Jeremy references at 09:04
• The SEO Arcade SOP on markdown companion files for AI crawlers
• Jeremy's older take on content cannibalization, which this episode partly overturns
• Unlinked mentions and footprint triangulation, on the mention side of the conversation
• Podcasting as PR: be the expert AI cites
• Three link builders on what still works

About the show

Unscripted SEO is hosted by Jeremy Rivera. The full interview archive lives at unscriptedseo.com.