The Role of Structured Data in Modern SEO [Zagreb SEO Summit 2025]

Editor's Notes

• #1: Agenda https://docs.google.com/document/d/1tuWv4FjeEmuH5QC_RqAXXYK_I90HgW-WrtOqL2fvANM/edit?tab=t.0
• #2: Get slides, links and 15% off Premium
• #6: Marked the introduction of semantic search in SERPs https://developers.google.com/search/blog/2012/10/rich-snippets-guidelines https://searchengineland.com/google-search-now-supports-microformats-and-adds-rich-snippets-to-search-results-19055
• #7: 2012
• #8: Connecting entities and keywords to intent and context Provided standards and structure for how entities are interpreted by crawlers
• #9: https://developers.google.com/search/blog/2012/10/rich-snippets-guidelines https://searchengineland.com/google-search-now-supports-microformats-and-adds-rich-snippets-to-search-results-19055
• #10: https://developers.google.com/search/blog/2012/10/rich-snippets-guidelines https://searchengineland.com/google-search-now-supports-microformats-and-adds-rich-snippets-to-search-results-19055
• #12: https://developers.google.com/search/docs/appearance/structured-data/search-gallery Brought order and interoperability to the web’s meaning layer
• #13: Structured data seeps to a lot of Google assets - it’s everywhere. Now they’re getting into emails with contextual ads
• #14: https://docs.google.com/spreadsheets/d/1R_7Wq7PdgQ2LJckbAjQIifq8gd-rDgJG-rYxh506yVk/edit?gid=0#gid=0
• #15: 2012-today
• #16: Isn’t rolling back As long as the native SERP exists, SS remains relevant
• #17: Chat Responses All similar, all text heavy
• #18: Requies SEOs to consider how our content can contribute to these ongoing dialogues rather than just answering isolated queries
• #20: Structured data supports discoverability in voice, video, documents, and beyond You’re not optimising for algorithms anymore - you’re optimising for understanding https://ipullrank.com/vector-embeddings-is-all-you-need
• #21: Structured data supports discoverability in voice, video, documents, and beyond You’re not optimising for algorithms anymore - you’re optimising for understanding https://ipullrank.com/vector-embeddings-is-all-you-need
• #22: https://blog.google/products/search/ai-mode-multimodal-search/ However, AI Overviews and zero-click features are reducing visibility
• #23: 2000 https://www.webdesignmuseum.org/gallery/google-2000
• #24: https://commons.wikimedia.org/wiki/File:Tesseract-1K.gif
• #25: Discoverability is the new success metric
• #26: Human input is declining
• #27: Using fan out query
• #28: They read, summarise, and vectorise content.
• #29: https://ipullrank.com/how-ai-mode-works
• #30: Don’t assume they’ll interpret everything unassisted Brands (and individuals) may develop their own “LLM memory banks” Structured data may be how your content lives inside someone else’s LLM Connecting with the consumer via these agents Agents right now are lazy so any saving or efficiency boost will be favoured
• #31: LLMs prioritize semantic clarity, structure, and formatting cues (headings, lists, bold text)
• #34: https://invisiblegraph.com/semantic-stories/does-structured-data-markup-influence-eeat/ https://x.com/JarnoVanDriel/status/1857038935290970608/photo/1 SEOs must feed the models — don’t assume they’ll interpret everything unassisted
• #35: Top ways to ensure your content performs well in Google's AI experiences on Search
• #37: Don’t assume they’ll interpret everything unassisted Brands (and individuals) may develop their own “LLM memory banks” Structured data may be how your content lives inside someone else’s LLM
• #38: Don’t assume they’ll interpret everything unassisted Brands (and individuals) may develop their own “LLM memory banks” Structured data may be how your content lives inside someone else’s LLM
• #40: Some may say “you’re showing a thief around your house”. No. This is more like an estate agent info pack.
• #42: https://x.com/levelsio/status/1905676526533652584
• #43: https://x.com/jeremyphoward/status/1905762648626798624 Go back to viewing a house
• #44: https://yoast.slack.com/archives/C07BMPEBJE7/p1746616337669359
• #45: LLMs prioritize semantic clarity, structure, and formatting cues (headings, lists, bold text)
• #46: Ignore it, fine. But what if you shouldn’t? It’s like buying a book without the synopsis in the back, knowing that’s where one would naturally look. That way you have it
• #47: 2000 https://www.webdesignmuseum.org/gallery/google-2000
• #49: Traditional methods, requiring individualized integrations and more manual oversight https://norahsakal.com/blog/mcp-vs-api-model-context-protocol-explained/
• #51: Unified interface for AI agents to dynamically interact with external data/tools https://www.linkedin.com/posts/whitebeth_modelcontextprotocol-ai-aiagents-activity-7329132096680009728-iUTI/ https://norahsakal.com/blog/mcp-vs-api-model-context-protocol-explained/
• #52: Unified interface for AI agents to dynamically interact with external data/tools https://modelcontextprotocol.io/introduction
• #53: 2012 - same as Knowledge graph November 25, 2024 (12 year gap, this is 8 weeks) https://www.webdesignmuseum.org/gallery/google-2000
• #54: Open project conceived and developed by R.V. Guha, creator of widely used web standards such as RSS, RDF and Schema.org
• #56: Every NLWeb instance is also an 𝐌𝐂𝐏 𝐬𝐞𝐫𝐯𝐞𝐫, which allows websites to make their content discoverable and accessible to agents via one core method, 𝘢𝘴𝘬, which is used to query a website with a question in natural language.
• #57: 2012-today (13 years) This is just under 6 months May 19th 2025,
• #58: https://www.linkedin.com/posts/zayarni_nlweb-rag-qdrant-activity-7330280592074264576-8-bd/
• #59: The year ahead https://commons.wikimedia.org/wiki/File:Tesseract-1K.gif
• #60: Ignore it, fine. But what if you shouldn’t? That way you have it
• #61: You’re not optimising for algorithms — you’re optimising for understanding Optimise content for discoverability, adding LLM parsing into consideration
• #62: Remember 15% off at link Follow me!