THE TOOLKIT

Appendix C — What the Evidence Says

Every tactic in the book, graded A to F against the published evidence. Where studies conflict, both are named rather than reconciled. Use this table to decide what to fund, what to trial with a measurement plan attached, and what to decline regardless of how confidently it is sold.

From Becoming the Answer by Jeremy Osborn · 347 words

Every tactic in this book, graded. Where studies conflict, both are named.

TacticEvidenceGrade
Server-side render primary contentVercel (1.3B crawler requests, 2024); searchVIU (23 crawlers, 69% cannot execute JS, 2025)A
Fix AI-crawler 404s and redirect chainsVercel: ~35% of ChatGPT and Claude fetches hit 404sA
Correct crawler access control by user agentFirst-party documentation from OpenAI, Anthropic, Perplexity, GoogleA
Product feed completeness and the eight Merchant Center attributesGoogle first-party specification, stated AI-surface purposeA
Never seed or sentiment-incentivize reviewsFTC rule effective October 2024; enforcement live 2026A
Report rolling averages onlySt. Gallen: run-to-run source overlap 0.34–0.42A
Entity consistency and WikidataDocumented grounding mechanism; strong correlational supportA−
/.well-known/ucp manifestGoogle first-party specification; Universal Cart live in the USA−
Earned third-party mentionsAhrefs: strongest correlate; Muck Rack: 82% of citations earnedB+
Digital PR aimed at cited outletsMuck Rack; third-party listicles ~80% of listicle citationsB+
Publishing prices and specificationsLargest controlled experiment: price presence among the strongest factors, in a synthetic testbedB+
Agent-completability testingAgentic traffic composition; 70% product and search routesB+
Organizing coverage around real question-form queriesWashington University: 6.8x activation on question-form queries (query form, not page formatting)B
Original data, statistics and named sourcesAcademic GEO study; partially challenged by C-SEO BenchB
Content recency and refresh cyclesSeer: 65% of AI bot hits target content under a year oldB
Passage self-containmentDocumented chunking mechanism; direct effect unmeasuredB−
YouTube presenceStrongest single correlate (r ≈ 0.74); causality unestablishedB−
Structural rewriting for chunkabilityLargest controlled experiment: negligible. C-SEO Bench: sometimes harmfulC
JSON-LD schema as a citation leverAhrefs difference-in-differences (n=1,885): −4.6% to +2.4%; Google: no special markup neededC−
Mass-generated permutation pagesContent volume is the weakest correlate measured (r ≈ 0.18)C−
llms.txtZero frontier-crawler fetches in a 900-domain, seven-month log studyD
Blocking Google-Extended for AI Overview controlGoogle documentation: does not affect Search inclusion or AI featuresD
Keyword stuffing−8%, the worst-performing method testedD
Manipulating AI recommendationsWorks, then degrades for everyone; plus regulatory exposureF

Back to the full contents of Becoming the Answer