Winners & Losers of the AI Disruption of Search

I'm humbled to be here this morning. Thank you very much to AirOps for flying me out to this very green room that looks like Green Lantern and the Riddler had a baby.

My name is Kevin Indig. I'm an advisor to fast growing tech companies, and I send my research and thoughts to over 26,000 people every week. I'll be honest: a couple of months ago I had a bit of an existential moment. For 15 years, SEO has shaped my career and, honestly, a little of my identity too. What I found is that AI completely changes the scoreboard for what works in SEO.

I felt grief. But grief is a signal for data, and the data shows the old scoreboard does not work the way it used to. So I set out to figure out what the new scoreboard looks like. Over the next 20 minutes I'll show you exactly that: six sectors where AI changed who the winners and losers are. I looked at over 20 months of my own user-behavior research, analyzed more than 1 million AI answers and over 100,000 AI citations, and spent about 30 hours of my life I won't get back.

1. AI Search

Search is at the heart of AI because it is the touchpoint most people have with it. There was a hope that AI would democratize search and give new and upcoming sites a chance. But the data shows the old dynasty kept its throne. Search rank has the highest correlation with citations in AI. If you rank number one in web search, you will probably get four times more citations than number 10.

There is some truth to the idea that SEO equals AI optimization, but it is not the same thing. SEO is the skeleton of the body that is AEO. And people keep their habits. In AI mode, the majority of people keep choosing the top result when they get a short list of products. AI short lists are a podium of one.

But one thing trumps rank, and that is trust. When people see an AI answer, the first filter they apply is trust: they check whether they trust the source or citation, and only then do they ask whether it actually answers their question. Trust is a scarce resource that is gaining value in this new AI world. In my research I watched a real person in AI mode hover over the second result, from a brand they trusted, and choose it over the brand that ranked on top, and this was a high-priced laptop purchase.

There is really no middle class in citation land. The top 10 domains in any vertical get almost 50% of all citations. If you are not in the top 10, you are fighting for scraps.

The losers. First, prompt tracking. When you run a prompt three times consecutively, only about 2.2% of citations remain. It is a probabilistic system with high variability. The average prompt is about five times longer than the average search keyword, so every prompt is essentially unique. Think about prompt tracking more like tracking the weather than tracking your inventory.

Traffic is the other loser. Pew Research looked at 900 US adults: when Google shows AI answers at the top, the click-through rate of the classic 10 blue links drops by 50%. Only about 1% of users click on citations in AI answers. These systems are not designed to send traffic. It is a visibility play, not a traffic play. The same holds for ChatGPT and AI mode. I will leave this section with one client example: they lost about 50% year over year in organic traffic, but gained 20% in conversions year over year. Purchase intent is not gone. People are still buying, they are just not clicking. They are looking at AI mentions and citations instead.

2. Content

Content is a sector where AI changed the unit economics. It is cheaper and faster than ever to create content, but creators and companies now compete with synthetic content on every platform. The bar for quality has risen sharply.

There are around 16 billion searches on Google every day, which sounds nice until you look at YouTube shorts, which get 12 times more. If Google searches were a month, YouTube shorts are a whole year. Video stretches into B2B too: more channels are crossing 500,000 subscribers, like Shopify, HubSpot, and Ahrefs, giving a direct line to customers and prospects that no algorithm can block.

YouTube is now the most-cited domain, more than Reddit, which makes it a core pillar if you want to be mentioned in AI answers. Video transcripts have the highest correlation with actually being mentioned in an AI answer. User-generated content like Reddit and YouTube also serves as a trust filter: when Google shows AI Overviews, clicks to Reddit and YouTube go up by two thirds, because people validate AI answers with answers from other humans.

The loser. Informational content. How-to and what-is keywords have lost 35 to 60% of click-through rate over the last 12 months. Informational content is the yellow pages of 2026. Answer platforms like Quora, Chegg, and Stack Overflow have lost over 60% of their traffic peak, and the stock does not look much better.

3. Ecommerce

You can vibe code Trello, but you cannot vibe code sneakers. AI can synthesize information, but it cannot synthesize physical products. It does, however, change how consumers shop. Salesforce found that 20% of shoppers in the last Black Friday and Cyber Monday season used AI. For context, healthcare is almost 20% of the US economy, and almost 20% of US energy comes from nuclear. Now 20% of online retail is touched by AI.

I looked at Shopify's earnings calls and projected their AI retail sales: they are close to $100 million in AI sales. That is only 0.1% of total GMV, but it is growing 34 times faster. This is a rocket coming toward us. The key behavior shift is that buyers are going direct to merchants. In travel research we found 80% of travelers go directly to airlines and hotels when they use AI, not to OTAs, which is bad news for Skyscanner and Tripadvisor.

The losers. The intermediaries. Affiliates like NerdWallet have lost over 70% of organic traffic over the last two years, while Chase Bank gained 10%. Being an affiliate in 2026 is like being a travel agent in 2008. At the same time, Google is becoming an ecommerce marketplace: almost every shopping query now shows a product carousel at the top, so shoppers can buy directly in Google without going to intermediaries.

4. SaaS

Everyone thinks AI is flattening SaaS. I think the opposite will happen. AI can compress the code layer, but it cannot compress organizational complexity. HubSpot coined the term inbound marketing, which is mostly SEO, and they lost about 70% of organic traffic in the last two years. But their stock doubled. Either SEO is not as efficient anymore, or they replaced it with something else, and it is the latter.

HubSpot replaced SEO with a media empire. They bought companies like The Hustle for $27 million, which came with 1.5 million newsletter subscribers, and built the HubSpot Media Network, now covering over 20 podcasts, 10 column writers, six YouTube channels, and several newsletters. A media empire is insurance against channel dependency: when algorithms limit your distribution, an owned media network keeps you in business. It is a trend, not a one-off. Companies like Okta and Ramp have built newsrooms with full-time journalists and data scientists. Traditional media is crumbling, but it is finding a new home in corporate media.

Why does it work? About 70% of citations in the software space go to vendors, and 70% of those go to blogs. It is about publishing insights nobody else has. Every B2B company should ask how to replace channel dependencies with owned audiences.

The loser. Aggressive AI content. It often works really well until it does not. Part of the problem is that SEO lives as an isolated discipline. Ask yourself: if you can create content within a minute, is it really that good? Time is a great heuristic for durable, sustainable content.

5. Measurement

Measurement is how we make sense of the world, and AI is breaking distribution. Last-click attribution is still stuck in 2006. Do not do that. Between 60 and 75% of marketers say their measurement is broken. Luckily, 50% of companies already invest in incrementality, and another 30% are planning to start. Attribution tells you who gets credit; incrementality tells you what actually works. Quick example: at my client Product Hunt, we added a last-updated date to a segment of pages and found a 3x increase in citations, a lever we can now deploy across the whole website.

The loser. Attribution. About 70% of AI traffic is attributed to direct traffic instead of AI referral traffic. It is 2026, and GA4 still does not have an AI channel. Primary data sources are getting more filtered too: Google Search Console data shows only 60% of clicks and 25% of impressions still arrive somewhere we can see and measure. If you are trying to track prompt performance in Search Console, you will see almost none of it, because it is filtered out for privacy.

6. Talent

Every three months I get a weird mix of excitement and fear because Anthropic launches a new feature and I wonder where my value in the world is. But I think the biggest danger of AI is not replacement. Dario Amodei, Anthropic's CEO, said we are about 6 to 12 months away from AI replacing software engineers. He said that five months ago, and the trend is not looking that way: software engineers are in higher demand than ever. Sam Altman, OpenAI's CEO, said with high confidence that customer support roles would be taken over by AI. That was eight months ago, and demand for customer support roles is now outpacing the broader job market. As long as these AI companies keep hiring, I will sleep well.

In all honesty, there are real efficiency gains from AI. It makes us about 25% faster and lets us do 12% more. The critical piece is that AI is good at taking direction and doing execution, but bad at giving direction. AI is good for execution, not strategy, and you can see it in the data: the most strategic tasks are where AI underperforms most. I am not an AI skeptic, I am a replacement skeptic. I benefit from AI: I build my own tools, like a title-testing engine and a tool that prioritizes topics for my clients.

The market reflects this. SEO jobs that demand AI skills pay over 25% more, and across tech, roles that require AI skills pay about 50% more. The art is navigating the frontier of what AI is good and bad at. The real downside is cognitive atrophy: over the last two decades our attention span dropped from two and a half minutes to about 40 seconds, while AI context windows grew by 4,000x, not 4,000%, but 4,000x. It is tempting to outsource thinking to AI, but the vicious circle is that doing so atrophies your cognitive abilities, which makes you outsource more. If you are worried about what social media does to society, you will have it coming for AI.

The new scoreboard

Prompt tracking matters, but we cannot dismiss rank tracking. There is still a lot to be said for classic SEO and its impact on AEO. We are moving from a world of traffic to a world of trust, from information to discovery, from intermediaries to sources, from answers to audiences, and from attention to judgment. My unifying theory is that AI changes humans more than it changes AI search, because it changes what we value: trust, authority, and judgment.

My prediction: in the next 24 months we will see a massive explosion of corporate media. Two drivers. First, legacy media is crumbling, so more journalists are joining the corporate world. Second, AI makes it easier to create products, so distribution and marketing get even more crowded, and companies will use media arms to fight for your trust and attention. I am very convicted in that. Thank you very much.