Lessons from the trenches of enterprise AI search

Search SEOul 2026.

We noticed something early in our business, working with category-leading teams. The difference is not tactics or headcount. It is a strong connection between the executive team and the search and product teams. We call this running the loop.

The loop is simple but not everybody makes it all the way through. You are not subscribing to tools. You are building a world with your data, and the best teams are very good at world-building. We can usually tell within 1-2 meetings which type of team we are working with. The search teams are generally pretty good. The question is always how much juice do they have.

The good news is that AI is opening new opportunities for search teams. Executive teams feel the pressure and are willing to listen in ways we have not seen before.

This talk covers how to capture that opening: the strategic questions executives are actually asking, the metrics that answer them across surfaces, and how to explain the technical concepts visually and in their language.

It also covers the measurement work underneath. There is no traditional versus AI search anymore, only the user experience. We are well past rank tracking. SERP features and chat features both carry actionable competitive insight, mentions and citations are a start, and most of your users will not be human. First party and third party data need proper joins to give full context, which is what unified visibility provides, and the warehouse only matters if the loop actually uses it.

Iteration count is the difference between the category leaders and the stragglers.

Transcript

1. Lessons from the trenches of enterprise AI search Search SEOul 2026 Ray Grieselhuber Founder and CEO, DemandSphere · Columbus and Tokyo
3. This talk will be a little different from my usual data-heavy talks heavy talks 4
4. We noticed something early in our business, working with category-leading teams 4
5. The key difference between category leaders and stragglers: Strong connection between the executive team and search and product teams. We call this ”running the loop.” 4
6. The loop is simple but not everybody makes it all the way through
7. You’re not subscribing to “tools.” Think of it as building a world with your data. 4
8. The best teams are really good at world-building. 4
9. Key components that affect the loop Executive The leadership team, ideally with a vision that extends all the way from corporate strategy to world-class performance in search. The data Product The team with write access to the digital experiences. Owned as a strategic asset, with continuously refreshed context and memory. AI search teams The experts in understanding the gap between AI search potential and present execution. 5
10. We can usually tell within 1-2 meetings which type of team we are working with 4
11. The search teams are generally pretty good. The question is, always: how much juice do they have? 4
12. The good news is that the impact of AI is opening new opportunities for search teams. Executive teams feel the pressure and are willing to listen to their search teams in new ways. 4
13. How to capture the opportunity: 1. Identify the strategic questions they care about 2. Use simple, strategic metrics across many surfaces 3. Explain the technical concepts visually 4. Speak their language when communicating what you need 4
14. The key strategic questions and what to measure question what the answer is made of what it draws on 01 What is my market in AI search? the full prompt and query set where the brand could appear SERP analytics · LLM visibility · prompt research 02 How big is that market? impression volume, prompt frequency, share of voice potential share of voice · market trends · GSC and GA4 03 How much of it do we control now? share of voice plus citation and mention rate, per platform share of voice · brand mentions · citations 04 Who are the main competitors? who ranks alongside us, and who the models recommend beside us competitor discovery · LLM co-occurrence 05 Who is winning and losing? citation share and AI Overview presence over time, not snapshots trend analysis · citation share · alerts 06 Who else is taking share? publishers, aggregators and review sites cited ahead of the brands citation source analysis · entity tracking 07 How does this turn into outcomes? managed services, consulting, or platform integration three execution tracks 08 What results should we expect? citation share, share of voice, crawl health, attribution measurable, not vanity
15. Be able to show in concrete terms using metrics they understand How to lose the room How to build credibility Frameworks, maturity models, a promise to investigate, and any answer that takes too long to assemble. Concrete, transparent, and relevant metrics. Not too many, focused on answering the key strategic questions they care about. Share of Voice (SoV) is the easiest, single metric to use as a translation layer between all of the data available in the world of SEO / GEO / AEO and the language that the leadership team uses. 9
16. Share of Voice (SoV) is easy to understand and it works across all surfaces
17. Some key concepts to explain well (following slides) 4
18. There is no traditional vs. AI search. It’s ALL AI search. The only question is the user experience
19. We are way past rank tracking. Measure the shape of the surfaces that predict user behavior.
20. Every SERP Feature is full of actionable and competitive insights
21. The same is true with Chat Features
22. Mentions and citations are great but there is a lot more
23. Most of your users will not be human
24. Building the unified, full-context view 4
25. Combine first party data with third-party data with proper joins to provide full context
26. Unified Visibility makes the connection between GSC data and SERP analytics
27. Building an operational data warehouse
28. Data ownership means the warehouse is used in the loop
29. Iteration count is the key difference between the category leaders and the stragglers 4
30. Speak the language of the executive teams Much of the work in SEO / AEO / GEO spans two different financial models. Know which one you’re requesting. Capex Opex Site level Content operation Platform, templates, rendering, structured data, architecture. Funded once, benefits every page, owned by product and engineering. A roadmap conversation, won six months before the work starts. Creation, optimization, refresh, coverage, feeds. Runs continuously, scales with headcount and budget, owned by marketing. A planning conversation, won every quarter. 24
31. What the loop output looks like
32. Come talk to us at our booth! 32
34. Thank you 감사합니다 Ray Grieselhuber · Founder and CEO, DemandSphere @raygrieselhuber 한국어 linkedin.com/in/raygrieselhuber DemandSphere in Korean speakerdeck.com/raygrieselhuber github.com/raygrieselhuber demandsphere.com/kr ray@demandsphere.com 31