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

Google's AI Overviews - The New Search

Google's AI Overviews - The New Search

Slides from my talk a the 2025 WAN-IFRA World News Media Congress where I spoke about the impact of Google's AI Overviews, how to optimise for them, and what the other chatbots have done to compete with search.

Avatar for Barry Adams

Barry Adams

May 06, 2025
Tweet

More Decks by Barry Adams

Other Decks in Marketing & SEO

Transcript

  1. Barry Adams ➢ Active in SEO since 1998 ➢ Specialist

    in SEO for News Publishers ➢ Newsletter: SEOforGoogleNews.com ➢ Co-founder of the News & Editorial SEO Summit
  2. AI Overviews are made by Generative AI • Using Google’s

    own ‘Gemini’ Large Language Model • ‘Multi-step reasoning capabilities’ • But... not 100% accurate; ➢ All LLMs ‘hallucinate’ – make things up ➢ It’s because they’re not ‘real’ AI ➢ LLMs are super-advanced word predictors; -‘Probabilistic autocomplete’
  3. LLM with Search + Grounding ‘Grounding’ = Retrieval Augmented Generation

    (RAG) ‘Grounding’ LLM Question Search engine Grounded answers with links
  4. AI Overviews always cite sources • AI Overviews have at

    least 1 cited source link; ➢ Every segment in an AI Overview will have a source link ➢ Average number of cited links: 6 -Sidebar on Desktop, carousel on Mobile ➢ Most cited links are pages ranked on the 1st page
  5. AI Overviews in Europe • Austria • Belgium • Germany

    • Ireland • Italy • Poland • Portugal • Spain • Switzerland • United Kingdom Currently active in:
  6. AI Overviews in Europe ‘The flurry of EU tech rules

    including the AI Act, Digital Services Act (DSA) and Digital Markets Act (DMA), “hold back the launch of innovative features” leading to concerns around product innovation. “The EU is behind when it comes to product innovation, and users in Europe will have a less good product experience.”’
  7. AI Overviews and News • Theory: AIO and QDF are

    mutually exclusive; ➢ QDF: Query Deserves Freshness ➢ QDF is a query/topic classifier that triggers a Top Stories box ➢ QDF != AIO
  8. Google Says… “…we aim to not show AI Overviews for

    hard news topics, where freshness and factuality are important” https://blog.google/products/search/ai-overviews-update-may-2024/
  9. Optimising for AI Overviews • Many cited sources in AI

    Overviews do not rank on the first page of Google’s regular results • AI Overviews cast a broader net than regular Google results https://ziptie.dev/blog/state-of-ai-overviews-5-google- aio-disruptions-found-after-analyzing-500k-queries/
  10. Content Segmentation / ‘Chunking’ • Divide your articles into concrete

    paragraphs/sections; ➢ Provide a strong subheading to introduce the section ➢ Phrase the subheading as a question • Each paragraph/section should have its own topical focus; ➢ Minimise the mention of other topics/entities in a section • Increases probability the page is used as a cited source; ➢ If the paragraph/section aligns with the user’s prompt query ➢ Taps into LLMs’ RAG/Grounding behaviours
  11. Key factors for showing in AI Overviews • Quality •

    Relevancy • Authority • Accuracy ‘Classic’ SEO signals
  12. AI Overviews = Lower CTR • Despite Google’s claims, AI

    Overviews send less traffic to websites; ➢ Users find the answer directly in the AI Overview • The good news..? ➢ AI seems to generate more search queries on Google
  13. Fighting against AI Overviews? • Prevent your content from being

    used to train Google’s Gemini LLM: User-agent: Googlebot-Extended Disallow: / • Probably a good idea; ➢ Prevents your copyrighted content from being used
  14. Fighting against AI Overviews? • Prevent your webpages from being

    linked as a source in an AI Overview: <meta name="robots" content="nosnippet" /> • Probably a bad idea; ➢ Also impacts presentation in all other Google surfaces
  15. Optimising for AI Search • LLMs need to crawl the

    web; ➢ Technical SEO principles for crawling and indexing • LLMs ingest content; ➢ On-page SEO principles for relevancy • LLMs focus on relevancy and brand signals; ➢ Off-page SEO principles for authority
  16. Key factors for showing up in AI Search • Quality

    • Relevancy • Authority • Accuracy ‘Classic’ SEO signals
  17. AI Search will not replace classic search • Google’s AI

    Overviews are an evolution; ➢ There will be some traffic loss ➢ Focus on building strong brand signals ➢ News is excluded from AIO (for now) • AI offers alternative ways to search; ➢ Market shares are small… so far ➢ Not yet worth focusing on ➢ But, be ready to pivot ➢ It’s still SEO