Karsten Weide

Following up on this post from yesterday - which has generated some strong public and private reactions. To understand the challenge here, it is helpful to have a sense of where AI Search is likely to wind up, and why this is a particularly difficult challenge for @Google I’m going to vastly oversimplify this: Start here: Traditional Search is a question with a SERP delivered with links, content and a load of paid advertising based on the content of that question. For example: “where is the best coffee shop near me?” Google has dominated this use case for decades and it’s a 200B+ annual revenue business. AI Search is a conversation that starts with a question that may seem similar to traditional search, but is different in 2 fundamental ways. First, the question contains much more memory and context, and 2nd - the AI Search product is specifically designed to continue the conversation and ultimately deliver a smaller set of better answers (or just one, executed by the AI). The same question asked of AI Search - “where is the best coffee shop near me?” Is transformed using memory and context into the following (massive oversimplification): The AI knows that I always travel with my dog. That I eat Gluten free. That I prefer Organic options. That I don’t like noisy environments because I’m usually working while I get coffee. My AI will almost certainly ask me follow up questions about whether I have the dog with me, am I with anyone, do I want coffee or am I also looking for gluten free snacks/food, etc. One can assume that the “query growth” @Google is talking about is at least in part driven by this dynamic. And yes, that kind of conversational “query growth” DOES ultimately lead to a more intent rich, qualified “moment” in the conversation - where a paid offer or ad can be presented. Conversational AI Search will ultimately consume search, because it is already a better product AND because that product gets better the more I use it…because..memory/context. So where does that lead us and why is it such a challenge in particular for Google? 1. The monetization model for conversational AI Search is not close to worked out - but the natural end state here is that a very qualified sponsored offer can be presented to the customer at the perfect moment, inside the conversation with disclosure that it is “paid/sponsored”. The primary challenge for anyone providing an AI search experience will be maintaining trust that the offer is the RIGHT offer, not the highest paying offer. (This is an entirely different post). As we move toward this future, AI search challengers like @OpenAI @sama @AnthropicAI @Muse etc can move aggressively toward an experience that is very lightly monetized with ads, or ad free in the short term, to win share (and the memory/context dynamic makes this share more durable) Google is in a pickle here, because every time they route a search from traditional SERP result into a conversational AI, they’re sending that user down a path that is far less likely to generate a paid click, and far more likely to provide a great consumer experience - which in simplest terms means they are choosing between share of consumer search in the future and dollars today. This, like so many economic problems is a kind of duration mismatch problem. That consumer is worth $2 to me on this search, and tens of thousands of future dollars and I have to balance the need for money today with the lifetime value of that searcher. If you believe (like I do) that AI Search/assistants will become the largest ad market ever created, every 1% of consumer search SHARE @Google cedes will be worth many billions of dollars of annually recurring, high margin revenue in the future. @sundarpichai @sama @satyanadella @finkd @alexandr_wang @thsottiaux @elonmusk Am I wrong about this?