AI Search Reality Check: Making Sense of Change with Imperfect Signals

Janaina explores how rapid AI changes have turned search into a moving target, creating pressure and uncertainty for SEOs. This talk focuses on cutting through the noise, redefining meaningful signals, and making smarter decisions with imperfect data in an AI-driven search space.

Transcript

1. AI Search Reality Check: Making Sense of Change with Imperfect Signals Janaina Barreto-Romero Oncrawl Speakerdeck.com/janainabr
2. We are in a discovery phase, and it’s been uncomfortable.
3. We’ve all been dealt a specific hand and that’s the game. ? ? ? ?
4. You can’t see all the cards. Nobody can.
5. You can’t see all the cards. Nobody can. Traffic is down 2 ♠ 2 ♠
6. You can’t see all the cards. Nobody can. Traffic is down 2 ♠ 2 ♠ Attribution is unclear 7 ♥ 7 ♥
7. You can’t see all the cards. Nobody can. Traffic is down 2 ♠ 2 ♠ Attribution is unclear 7 ♥ 7 ♥ Unproven tools 5 ♣ 5 ♣
8. Not everyone has a Royal Flush. Impressions stable (for now). Clicks down. Sessions falling. And your exec needs an answer now. A ♦ A ♦ ♦
9. You don’t know what you don’t know.
10. The Noise
11. ♠ ♥ ♦ ♣ SEO is dead ☠
12. ♠ ♥ ♦ ♣ AEO/GEO is everything
13. ♠ ♥ ♦ ♣ LinkedIn says: “This is how it actually works!”
14. ♠ ♥ ♦ ♣ Click here for my $100K a month revenue generating AI strategy $$$
15. Half the table is bluffing.
16. The Truth
17. ♠ ♥ ♦ ♣ LLMs introduced a new channel of discovery →
18. ♠ ♥ ♦ ♣ Serp real estate is shrinking.
19. ♠ ♥ ♦ ♣ Traffic that used to reach your pages is resolved at the SERP level.
20. ♠ ♥ ♦ ♣ There’s a new way for content to be found. That's an opportunity.
21. ♠ ♥ ♦ ♣ There are new bots to investigate →
22. ♠ ♥ ♣ ♠ ChatGPT-User ♠ OAI-Searchbot ♠ Perplexity-User ♠ Claude User ♠ DuckAssistBot ♠ CCBot ♠ MistralAI-User And the list goes on… ♦
23. ♠ ♥ ♦ ♣ Crawlability still matters →
24. ♠ ♥ ♦ ♣ If a bot can't reach your page, it can't cite it.
25. ♠ ♥ ♦ ♣ The same technical fundamentals that apply in traditional search matter for LLMs.
26. ♠ ♥ ♦ ♣ Indexability still matters (to an extent) →
27. ♠ ♥ ♦ ♣ LLMs do rely on search index for retrieval, it’s one of it’s sources.
28. ♠ ♥ ♦ ♣ Adaptability as an SEO is key →
29. ♠ ♥ ♦ ♣ You need to be testing.
30. ♠ ♥ ♦ ♣ You have to let go of the old measurement rules.
31. ♠ ♥ ♦ ♣ Not playing is a risk, start now even small.
32. Reading Your Hand
33. A good player doesn't need a perfect hand. They need to know exactly what they're holding.
34. Play the cards in your hand A ♠ ♠ Log file data K ♥ ♥ Query fan-out Q ♦ ♦ Brand presence J ♣ ♣ Technical debt 10 ♠ ♠ Revenue alignment
35. ♠ ♥ ♦ ♣ Clicks are down. Is revenue?
36. ♠ ♥ ♦ ♣ Practical check for revenue. Keep it simple.
37. ♠ ♥ ♦ ♣ 10 ♠ Use your tool of choice to visualize 10 ♠
38. ♠ ♥ ♦ ♣ K ♥ Ingest revenue and traffic data K ♥ 10 ♠ Use your tool of choice to visualize 10 ♠
39. ♠ ♥ ♦ ♣ A ♠ Cross analysis of traffic and revenue on important pages A ♠ K ♥ Ingest revenue and traffic data K ♥ 10 ♠ Use your tool of choice to visualize 10 ♠
40. ♠ ♥ ♦ ♣ Technical debt is a visibility risk.
41. ♠ ♥ ♦ ♣ Practical questions to address technical blockers.
42. ♠ ♥ ♦ ♣ A ♠ What's blocking AI visibility right now? A ♠ K ♥ Are your key pages readable to AI bots? K ♥ 10 ♠ Are your key pages crawlable to AI bots? 10 ♠
43. ♠ ♥ ♦ ♣ Brand Presence What are people saying about you? What are LLM’s saying about you? What are you saying about yourself?
45. ♠ ♥ ♦ ♣ Sentiment Check K ♥ Test with search enabled K ♥ 10 ♠ Test the model training data 10 ♠
46. ♠ ♥ ♦ ♣ Prompt LLM Brand Mentioned Sources cited Sentiment ChatGPT Perplexity Gemini Sentiment Check
47. ♠ ♥ ♦ ♣ Why Does Query Fan Out Even Matter?
48. ♠ ♥ ♦ ♣ QFO: Content Audit A ♠ Use a tool that leverages an LLM to analyze your content A ♠ K ♥ Run manual checks across multiple LLMs K ♥ 10 ♠ Investigate the fan-out 10 ♠
49. ♠ ♥ ♦ ♣ Log Files
50. ♠ ♥ ♦ ♣ Log Files Bytes
51. ♠ ♥ ♦ ♣ User bots
52. ♠ ♥ ♦ ♣ Log Files
53. ♠ ♥ ♦ ♣ Log Files
54. Go All In
55. You go all in on the best information available.
56. ♠ ♥ ♦ ♣ B2B Client Case →
57. ♠ ♥ ♦ ♣ Client saw a 25% drop in traffic
58. ♠ ♥ ♦ ♣ Execs wanted to know what was happening
59. A ♠ ♠ Log file data K ♥ ♥ Revenue alignment Q ♦ ♦ Query fan-out J ♣ ♣ Brand Presence 10 ♠ ♠ Technical Debt What cards did they have?
60. K ♥ ♥ Revenue alignment Revenue was a strong positioning with execs
61. Prioritised technical fixes 10 ♠ ♠ Technical Debt
62. A A The results → The traffic that left wasn’t the traffic that mattered. → The traffic was more qualified. → Showed management there was no dip in revenue. B2B Client Case A ♦ A ♦ ♦
63. You don’t need to fold.
64. ♠ ♥ ♣ Scan to connect linkedin.com/in/janaina-br/ THANK YOU! 10 ♠ Let’s chat at the Oncrawl booth! 10 ♠