Vivan Marwaha - SEO in a Post-Google World

1. SEO in a post-Google world
2. Introductions Vivan Marwaha Profound Content Marketing Lead at Profound AI marketing platform that helps companies win more customers. 1,100+ enterprise brands, including a third of the Fortune 100 (Comcast, Walmart, US Bank) and companies like Ramp, Figma and MongoDB.
3. For 20 years, we focused on: The query The click The buyer Was a keyword with a volume you could look up Was a referrer you could count Was a person reading your page Now a prompt: long, personal, multi-turn Now a direct visit or a Google search the AI never gets credit for Now an agent with a budget and three sites to check
4. AI search is a bit different, and growing fast 900M 1B 1B weekly ChatGPT users monthly Gemini app users monthly Google AI Mode users
5. …even on Google 2× 25 yrs AI Mode queries more than double every quarter since Google last rebuilt the search box. It did it for AI Mode
6. And the newest users are not people Two personal agents launched this fall 10M+ $200+ $10B Muse users in its first six weeks, 1 in 4 of them daily average purchase for a quarter of the people who shop through Muse Instinctʼs valuation in September. It is a phone number you text
7. Agents are already on your site 56% 1 in 10 of web traffic is bots. A shopper visits 5 sites; an agent visits 5,000 US online retail dollars spent through an agent by 2030, the base case. 1 in 5 is the high case
8. 01 The query People donʼt search in keywords anymore
9. 19.9% AI search conversations with commercial intent
10. Commercial conversations grew fastest Growth in weekly ChatGPT conversation volume, June 2025–June 2026 Commercial +119% Informational Generative +63% +37%
11. And people trust these models 58% 72% use ChatGPT as their primary model rank ChatGPT among their most trusted models
12. AI has reshaped the buying journey
13. The old purchase journey
14. People turn to AI in every stage Stage of the first shopping prompt Discovery 41.2% Consideration 12.4% Evaluation 24.0% No clear stage 22.5%
15. Most people arrive undecided 60% 19% name no brand in their first prompt come with a specific product in mind
16. 1 in 3 turns in a shopping conversation recommend a brand
17. 3.7 brands in the average recommendation
18. 16% responses recommend a clear top pick
19. Recommendations differ by industry Share of responses that mention a brand Travel 36.1% Retail 35.1% Motor vehicles Arts & entertainment 22.7% 16.3%
20. There isnʼt just one purchase journey Discovery Discovery Consideration Consideration Evaluation Browsing Evaluation Episode ends
21. Three behavioral patterns Profound research / Stage transitions in multi-turn conversations
22. Discovery trends 34% of discovery stays in discovery; only 10% move to consideration
23. Discovery queries Budget word ”affordable” 13.3% “Best X for Y” 5.3% Time constraint 4.0%
24. Consideration trends 11% of consideration turns lead to web browsing; the highest of any stage
25. Consideration queries “Compared to / with” 10.7% “A vs. B” 8.3% “X than” (better, cheaper) 2.1% “Alternatives to” 1.9%
26. Evaluation trends 40% of evaluation turns are followed by more evaluations
27. Evaluation queries Budget word ”affordable” 14.7% Exact price 12.2% Size constraint 5.1% Power/performance constraint 4.9%
28. Prompts follow a few templates Share of conversations containing each pattern “Best X for Y” 3.6% Peaks in discovery: 5.3% “Compared to / with” 2.3% Peaks in consideration: 10.7% “A vs. B” 1.5% Peaks in consideration: 8.3% “Alternatives to” 1.2% Peaks in consideration: 1.9% “X than” (better, cheaper) 0.9% Peaks in consideration: 2.1% “Top-N ranking” 0.7% Only in discovery: 1.0% Profound research / Share of all conversations
29. Price limits are mostly about being affordable 3% 12% set a ceiling such as under “$X” use budget working such “cheap” or “affordable”
30. 02 The click Barely anyone clicks, but the purchase still happens
31. AI-started journeys reach checkout sooner Median days from entry to checkout 5.2 days 6.7 days 6.8 days 8.5 days AI conversation Brand site Search Marketplace 1 week
32. 1 in 5 AI-started journeys end in a purchase Shares of episodes ending in transactions 21.4% 23.5% 17.1% 24.5% AI conversation Brand site Search Marketplace
33. AI purchases are hiding in plain sight
34. Barely anyone clicks on a link <2% 17% tagged AI-referral within the same journey visit a named or cited brand site
35. AI conversations are rarely the last step Last step before purchase site, share of AI usersʼ transactions Search 14.4% Another brand or merchant site 10.3% Social or forum 9.4% Google AI Overview/AI Mode 4.5% AI conversation ~1%
36. Google is the next stop Immediate next page after a shopping conversation google.com 29.0% youtube.com 2.7% facebook.com 0.9% amazon.com 0.9%
37. Hidden AI: purchases that follow a conversation Purchases by AI users 13.5% 28.6% Prior 7 days Prior 30 days
38. Hidden AI: purchases that follow a conversation Purchases by AI shoppers 23.7% 51.4% Prior 7 days Prior 30 days
39. 03 The buyer The click stops being human
40. Everyday T-Shirt $45 Size Small Medium Large Add to cart X-Large
41. Everyday T-Shirt $45 Size Small Medium Large Add to cart X-Large
42. Muse and Instinct are the leading edge of this shift
43. Josh needed sunglasses
52. One business made money. Another didnʼt.
53. Every single product, brand, and service will be subject to this level of evaluation from agents in the future
54. What makes them different Proactive Action-ready Context-loaded It acts before you ask It works on its own computer It already knows you
55. Muse The personal agent from
56. How Muse searches
57. How Muse searches
58. Museʼs written instructions for shopping
59. shopping / SKILL.md “You should always aim to save the user money.”
60. shopping / SKILL.md “default to searching a maximum of 3 merchant sites”
61. shopping / SKILL.md “ensure the top 5 [products] are the highest quality (…from well-known merchants/websites).”
62. How Muse does the shopping
63. “Whatʼs the best frying pan?”
64. This is the full task Museʼs browser got The user asked to find the best frying pan. Find real purchasable products online for: best frying pan for someone starting to cook, wanting a minimal set of nicer buy-it-for-life gear. Preserve these requirements: two angles worth covering — (a) one stainless steel tri-ply workhorse skillet (10-12 inch) that sears well and lasts forever, from well-reviewed brands like All-Clad, Made In, Tramontina, Misen, Cuisinart; (b) one good nonstick egg/omelet pan (8-10 inch) with durable coating, from brands like OXO, Made In, T-fal, Tramontina. Default to searching a maximum of 3 merchant sites and a maximum of 10 products in total …
65. Muse used the userʼs context
66. Muse chooses the size
67. Muse chooses the material
68. Muse turned one pan into two
69. Muse picked the brands
70. How Muse searches Observed workflow in recorded Muse shopping tasks · September 2026
71. Muse pulls from its browser predominantly but also uses a catalog Meta Catalog 60 from ~30 merchant sites Browser from 2 merchant sites In the answer 6 product cards 4 4 from the browser 2 from the catalog
72. Goes to your site and only sees the task
74. Agentic visits are growing rapidly Unverified meta-webindexer requests rose 1.6x as much on US and Canada sites as elsewhere
75. Make the product easy to evaluate and buy Publish the product facts and limitations Keep price, stock, and delivery conditions current Get into catalogs and product feeds where revelant Test if agents can buy on and use your site
76. Your product needs to be positioned for agentic attention
77. Make yourself legible to the agent
78. Until now, every journey started with Google. Now, it starts with a prompt