Upgrade to Pro — share decks privately, control downloads, hide ads and more …

Content Strategy for AI Search: What to do now?

Avatar for Matthew Brown Matthew Brown
September 24, 2025
0
4

Content Strategy for AI Search: What to do now?

A framework for understanding what LLMs actually do, and how Google has a myriad of online systems for understanding which content to surface in LLM answers, in contrast to where other LLMs (like ChatGPT) have significant shortcomings.

This deck covers four pillars of establishing visibility with current tools, using brand, personas, prompt generation, and tracking. It also covers things that might need some time to solidify before investing, and a warning about scaling content efforts using AI.

Presented at SMX London on 9/19/25 by Matthew Brown (https://mjblabs.com)

Avatar for Matthew Brown

Matthew Brown

September 24, 2025
Tweet

More Decks by Matthew Brown

See All by Matthew Brown
Do Core Web Vitals Actually Matter for SEO?
Avatar for Matthew Brown mjbpdx
0
200

Featured

See All Featured
BBQ
Avatar for Matthew Crist matthewcrist
89
9.8k
Side Projects
Avatar for Sacha Greif sachag
455
43k
I Don’t Have Time: Getting Over the Fear to Launch Your Podcast
Avatar for Joe Casabona jcasabona
33
2.4k
Automating Front-end Workflow
Avatar for Addy Osmani addyosmani
1371
200k
Balancing Empowerment & Direction
Avatar for Lara Hogan lara
4
660
The Cost Of JavaScript in 2023
Avatar for Addy Osmani addyosmani
53
8.9k
Distributed Sagas: A Protocol for Coordinating Microservices
Avatar for Caitie McCaffrey caitiem20
333
22k
Helping Users Find Their Own Way: Creating Modern Search Experiences
Avatar for Dan Newman danielanewman
30
2.9k
Why You Should Never Use an ORM
Avatar for John Nunemaker jnunemaker
PRO
59
9.5k
Writing Fast Ruby
Avatar for Erik Berlin sferik
628
62k
ReactJS: Keep Simple. Everything can be a component!
Avatar for Pedro Nauck pedronauck
667
120k
Reflections from 52 weeks, 52 projects
Avatar for Jefferson Lam jeffersonlam
352
21k

Transcript

  1. Content Strategy for AI Search: What to do now? Matthew

    Brown SMX London 2025

  2. About Me • I specialize in technical SEO and Product

    Management. • I primarily work with news and media, B2B, and social networks • I constantly think about our AI/SEO future too!

  3. Start with the right mindset The most important developments in

    SEO in the last two years weren’t even about AI!

  4. The Google Algorithm Leak

  5. The Google Exploit Data

  6. A terrifying little peek into Google search systems

  7. Hot take: They’re not lying to us

  8. So that’s why everyone is building a browser! Optimize for

    Google – The other AI engines are catching up.

  9. Google Leak Part II: The Entity Revealed This is crucial

    SEO work to understand. It’s the roadmap into how Google is building an AI search moat.

  10. Knowledge Graph > Link Graph https://1492.vision/research/entities-everywhere- knowledge-graph-google

  11. None

  12. This 2013 screenshot still works as a good SEO strategy

    today It’s how you are represented as a topical entity that matters This is what vector embeddings and semantic relationships get at

  13. From content math to personalization https://1492.vision/research/entities-everywhere- knowledge-graph-google

  14. The SEO goal: Ensure your content semantically aligns with the

    entities and relational factors that LLMs are focused on.

  15. https://ipullrank.com/ai-search- manual/quick-start-guide This is what Mike is talking about: How

    to *engineer relevancy* in these new visibility surfaces. Concepts: • Embeddings • Cosine Similarity • Passage Retrieval • Query fan outs

  16. The SEO goal: Ensure your content matches user intent and

    then back that up with strong user engagement metrics.

  17. https://www.airops.com/blog/webinar- mark-williams-cook This is what Mark is talking about: GEO

    tactics are just SEO tactics when you boil them down. Concepts: • Search behavior has changed but SEO strategy hasn’t (much) • We know Google is going to use a whole lot of signals other than link popularity and content relevance. • Offsite signals matter more now. You’re not just your website.

  18. Things I’d do right now: Brand Personas Prompts Content +

    Architecture

  19. Brand Sculpting If your brand isn’t represented in a subtopic

    you care about, there’s work to be done.

  20. Brand Sculpting Double check sources/grounding to see where you need

    visibility changes.

  21. Build Personas Personalization is going to be the biggest differentiator

    in AI search

  22. Tools I like for building personas: Gumshoe

  23. Tools I like for building personas: AirOps

  24. Prompt Generation

  25. Prompt Generation Tools I like for prompt generation: AlsoAsked.com

  26. Prompt Deserves Grounding: Pattern matching

  27. Prompt Tracking Tools I like for prompt tracking: Peec.ai

  28. What am I doing with this data?

  29. Build: shiny new HTML • Optimized CSS, especially header tags.

    • Clean <head> tags and divs • Use simple HTML elements like tables

  30. Build: Structured Data Don’t think of these as search ranking

    factors. Content in structured data may be parsed and help “preload the dice”

  31. Build: Topical Architecture • Use that fancy embedding math to

    reorganize site • Organize first by user intent/engagement • Then create topical hubs that link to a SMALL set of query fan outs. • Don’t boil the ocean: The less pages the better

  32. Build: Like you’re a Knowledge Graph • Use tools like

    GraphRAG to move a step beyond vector similarity • This is undoubtedly where Google already heading (or is there). • The short version: If there’s an entity you’re missing, it needs a hub page or a passage on the most semantically- related page

  33. Build: Recency Quick hack to do this: Track AI bots

    in server logs and establish a candidate list of URLs to refresh.

  34. Build: Audience Loyalty • Email newsletters (Exclusives!) • Member-only content

    access • Early product/feature access • Community participation opportunities • Direct engagement with experts • Personalized content experiences This is a great way your build your brand across channels. The goal is positive sentiment in the right places.

  35. Audience Research Tools: Sparktoro

  36. Build: multimodal content

  37. Things I’d wait out (or run experiments)!

  38. Citation building: Mixed results so far! What happens with everyone

    does citation building with old school SEO tactics?

  39. Query fan outs are good data – but need testing

    Source: Qforia

  40. What happens when it’s all passages? Search is going conversational.

    Passage spam is old SEO.

  41. Track new KPIs, but be VERY careful what you report

    on Can these be tracked against user engagement?

  42. A thing I’d be careful with

  43. Scaling Content with AI 2500 pages, 40% AI summaries, unique

    content scaled.

  44. Scaling Content with AI 500 pages, 15% AI summaries, unique

    content per page.

  45. Just when you think you’ve got it..

  46. None

  47. Elevator Pitch for that pesky exec Search Behavior is changing

    LLM adoption means we can’t just focus on keywords and pages. We have to research LLM usage and tailor our content approach We’re not just a website Our brand is an entity that lives across channels, and we need to be present in the right place for the right topics. LLMs might end up as visibility layers on search engines. Good SEO practices are still justified. SEO still matters

  48. It’s gonna be great!

  49. Questions? Email: [email protected] LinkedIn: MatthewBrownPDX