Homer's Guide to AI, Differentiation, Disambiguation, and Duff

Grant Simmons, formerly of Homes.com, uses humor and sharp examples to explain one of the hardest problems in AI search right now, how systems decide who or what you actually are.

As AI models pull signals from across the web, brands and entities that are not clearly differentiated get blurred, misattributed, or ignored. Grant breaks down how disambiguation really works, why many sites are unintentionally training AI to confuse them with competitors, and how small structural mistakes lead to big visibility losses. The session mixes laugh out loud examples with practical lessons drawn from years of large scale SEO and platform work.

You will leave with a clearer understanding of how AI interprets entities, how to reduce ambiguity around your brand, and how to stop sending mixed signals that undermine trust. It is educational, irreverent, and grounded in real world experience, with just enough Duff to keep it honest.

Transcript

1. Homer’s Guide to AI, Differentiation, Disambiguation, and Duff
2. Hi! I’m Grant Simmons
3. Hi! I’m Grant Simmons
4. Oh. And I’m British
5. @simmonet /in/simmonet Feel free to connect
6. Let’s Talk About AI (LLMs)
7. What Do We Know About AI?
8. AI Doesn’t Browse. It Picks.
9. Yes. This is a Simpson’s Themed Presentation About Serious Topics
10. Our (Serious) Agenda
11. But... Before We Start...
12. GE
13. We’re Going to Be Calling it GEO GE
14. And... Good SEO is Good GEO GE
15. John Said So
16. And Oogle Never Gets it Wrong
17. Get Picked AGENDA Be Understood Measure It
18. And FYI... We’re All Learning
19. Get Picked
20. AI Answers are Built From Chunks, Not Pages
21. The Winning Chunk is the “Liftable” Snippet
22. “Liftable” = Easily Extractable Snippet
23. The System Picks a Few Chunks Not the Whole Banana
24. Only Part of Your (Ideal) Brand Story
25. Study: Short, Specific Passages Win Disproportionate Attention Source: DejanSEO
26. If it’s Not Easily Extractable Your “About Page” Doesn’t Matter
27. There Are No Absolutes. Even This Statement.
28. “If LLM Can’t Lift It, They Can’t Cite It.”
29. The Process Why a Chunk is Picked
30. Picked Chunks are Self-contained.
31. Picked Chunks are Specific - (Numbers > “Vibes”)
32. Picked Chunks are Verifiable - (Nearby Proof)
33. Clear Nouns. Clear Referents (Not: “It Does This”) Picked Chunks Reduce Ambiguity
34. The Prompt’s “Job-to-be-done” Picked Chunks Align With Intent
35. Consensus Sources Can Beat “Best Writing”
36. Consensus vs Originality / Contrarianism
37. Good SEO is Good GEO. In to Win.
38. How to Get Picked
39. One Paragraph. One Idea. Design for Extraction
40. Front-load Your Answer. Don’t Bury It.
41. Requirements > Methodology > Outcomes > Caveats Turn Claims Into Mini-specs
42. “Best” > “Best for X” Replace Adjectives With Caveats
43. Links, Citations, Primary Data Ensure Proof is Close to the Claim
44. Use Comparison Grids (LLMs Love Decision Structure)
45. Present for Humans
46. Write “Known for” Statements repeat them everywhere.
47. Write “Known for” Statements repeat them everywhere. “Homer is known for living in an imaginary town called Springfield, his love of eating donuts and his excessive consumption of Duff beer at Moe’s Tavern”
48. Create “Citation Magnets” Tools, Templates, Original Assets
49. I Call Them “Lead Magnets” (for Humans)
50. Don’t Answer the Same Thing Twice Differently Avoid Redundancy
51. Not Hidden in Tabs/Accordion Purgatory Make Key Content Accessible
52. “Liftable Beats Lovable.”
53. Accuracy & Alignment
54. They Lose to Confusion Most Brands Don’t Lose to Competitors
55. If Your Entity Isn’t Clear...
56. You’re Going to Be “Averaged” • Described like a competitor • Wrong geo/industry/audience • “You do everything” positioning • Misattributed information • Founders, pricing, capabilities
57. Disambiguation is Not Branding
58. It’s Identity Hygiene. Disambiguation is Not Branding
59. Who You Are. What You Do. (And Contrary) Build a Canonical Entity Page
60. About Page on Steroids
61. Document What AI Gets Wrong
62. Waikay J
63. Fix “Collisions” with Qualifiers
64. Write Simply for Humans (AI is Dumber)
65. Consistency Across the Web Beats Clever Site Copy
66. Schema Can Help Clarify. But Test.
67. SameAs Schema (scan to learn)
68. Measurement
69. Rankings Won’t Show Misattribution
70. Mentions, Citations, Share-of-Prompt, Accuracy.
71. “Share-of-Prompt?” Yes. It’s a Thing.
72. Oops
73. Like SEO isn’t About Optimizing Search Engines
74. Wrapping Up
75. “Good SEO is good GEO” • Fundamentals still power selection: relevance, helpfulness, structure, experience • “GEO” is mostly SEO + selection mechanics + measurement • Key nuance: the format of retrieval/citation is changing even if the inputs rhyme
76. What Gets Picked • Chunks not pages • “Liftable” snippets *Caveat. Caveat. Caveat.
77. Why it Gets Picked “Chunk-first” Publishing • Self-contained • Short, specific paragraphs win* • Verifiable • Clarity • Aligned-intent • Consensus* *Caveat. Caveat. Caveat.
78. How to Get Picked Design for Extraction • One paragraph, one idea • Claims > mini specs • Replace adjectives with caveats* • Keep proof close • Write “Known for” statements *”Best for caveats”
79. Accuracy & Alignment • Don’t lose to confusion • Don’t be average • Identity hygiene • Track / fix hallucinations / errosr • Consistency beats clever • Schema (why not?)
80. Measurement Measure what you can J • Mentions • Citations • Share of Prompt • Accuracy • Prompt tracking in moderation* * Directional
81. “Trying is just the first step toward failure.” ~ Homer
82. Do or do not. There is no try.
83. we are all learning
84. Thank You!
85. @simmonet /in/simmonet Feel free to connect waikay.io (free trial)