The Visit That Never Comes

Search, Ads, AI and the Next Transformation We Are Already Living Through

Issue 290, November 12, 2026

Think about the last time you needed a new pair of running shoes, a dishwasher or earbuds. Not long ago, the routine went something like this. You typed a few words into Google and scrolled past the ads at the top. You opened six tabs, read two reviews that contradicted each other, went to Amazon and scrolled through a page of sponsored listings. Somewhere in there, an email arrived from a retailer you bought from three years ago, reminding you there was a sale. Eventually, more worn down than convinced, you bought something.

Now a lot of us just ask AI. We type or say the question to ChatGPT, Gemini, Claude or the AI summary sitting at the top of the Google results, and we get an answer without going any further in the search results. Three options, a sentence on each and maybe a recommendation. Plenty of people stop right there. They never open the six tabs, never see the sponsored listings and never land on the company’s website at all.

This is a big shift. For more than twenty years, organizations have built their marketing around a simple assumption: a person with a question would come looking, and if you did the work, they would find their way to you. Search engine optimization, paid search, content marketing, email lists and the whole advertising machine were built to win the eyes and that visit. People are still asking the questions, more than ever, in fact. What’s disappearing is the visit, as we move to simpler, faster and seemingly more efficient ways of getting answers.

The ingrained behaviors and practices we embraced to fulfill our needs and wants have, in many ways, stopped, and the visit has gone with them.

What we are experiencing right now- the shifts, the new opportunities, the blocking, the lawsuits, the scrambling, and the deal-making- isn’t the end of the story, because ultimately consumers will drive what comes next and what form it takes. But if history is any guide, we are in the early and noisy part of a huge transformation many of us have lived through before, just in different ways.

Twenty Years of Building for the Click

Companies of every size spent years and a lot of money learning how to rank. They hired SEO specialists and agencies, rewrote web pages around keywords, built content calendars to feed the search engines, chased backlinks and watched their rankings the way some people watch the stock market. Marketing teams were organized around it and budgets were defended with it. Entire careers were built on knowing how Google decided what showed up on page one, or at least on knowing more about it than the person in the next office.

And for a long time, it worked. The formula was reliable enough that people trusted it. Do the work, rank well, get the traffic and convert some of it. The dashboards showed the results every month. That kind of track record builds confidence, and as I wrote last week, confidence built on a story that has held together for years is very hard to question, even after the conditions underneath it have changed.

Every one of those investments rested on the same assumption: that somewhere at the end of the effort, a person would arrive on your page.

What the Numbers Already Say

In 2025, Pew Research Center looked at the actual browsing activity of 900 American adults who agreed to share it. When Google showed an AI summary at the top of the results, people clicked on a regular search result in 8 percent of visits. When there was no summary, they clicked 15 percent of the time, nearly twice as often. Clicks on the links inside the AI summary itself were rare, about 1 percent. People were also more likely to end their browsing session altogether after seeing a summary. They got what they needed and left.

Search itself, though, isn’t shrinking. In February 2024, Gartner predicted that traditional search engine volume would drop 25 percent by 2026 as people moved to AI chatbots. In late October, reporting its third-quarter results, Alphabet said the opposite happened. Sundar Pichai said Google’s year-over-year growth in searches was the highest it had been in five years, and he credited AI for it. People are asking more questions than ever. They are just getting their answers without going anywhere.

That is starting to show up in the money. Google’s search revenue was $63.35 billion in the third quarter, still up 12 percent from a year earlier, so this isn’t a collapse. But the growth rate slowed from roughly 17 percent the quarter before. On the earnings call, Google’s chief business officer, Philipp Schindler, named currency swings as the biggest reason. He also said the fast growth in AI Mode was having an effect, because Google is earning less from those AI answers than from traditional search results so far. It is the first time the company has openly acknowledged that answering the question right there on the page comes at a cost.

The Fight Over the Shelf

The same pressure is now reaching retail, and Amazon is where it is most visible.

On September 8, Meta launched an AI assistant called Muse that can book, buy and fill out forms on a person’s behalf. Within two weeks, Amazon blocked it. Amazon said Meta never got permission, that the agent didn’t identify itself while it browsed and that it appeared to be storing customers’ account credentials, which Meta denies. Amazon has gone after other agents too. It sued Perplexity over its Comet browser’s shopping assistant, won a court order blocking it in March and then saw that order reversed on appeal in August. That fight is still going.

Some of Amazon’s concerns are legitimate and should be taken seriously. Nobody should want an unidentified piece of software holding their passwords. But there is more at stake than security. When you scroll Amazon yourself, you see sponsored products, and Amazon’s advertising business brought in roughly $68 billion last year, according to Forbes. When an AI agent shops for you, it doesn’t scroll. It doesn’t see the sponsored listings, and it doesn’t care how much a brand paid to sit at the top of the page. It simply finds what fits the request.

So the blocking is, at least in part, a protection mechanism, an established business model defending the place where it makes its money. That is a completely understandable response, and one we have seen before.

We Have Seen This Before

If you were around in 1999, you probably remember Napster. It launched in June of that year and let anyone share music files with anyone else, for free. Within six months the Recording Industry Association of America, on behalf of the major labels, had sued it. Metallica sued it as well. The courts ordered it to stop sharing copyrighted music, and in July 2001 Napster shut down. In September 2003, the industry went further and began suing individual people, naming 261 music fans in its first round of lawsuits.

The industry won nearly every one of those fights in court, and it still lost its customers.

Consumers had already seen what was possible, and they weren’t going back to buying a whole CD to get the one song they wanted. When Napster closed, other file-sharing services took its place. In April 2003, months before the lawsuits against fans began, Apple opened the iTunes Music Store and sold songs for 99 cents each, with all five major labels on board, because by then the labels needed somewhere to go. Then came streaming. By 2016, streaming made up 51 percent of all recorded music revenue in the United States, up from 9 percent just five years earlier. And the industry’s total revenue that year was still only about half of what it had been in 1999.

It is a pattern I come back to often. In The Drift That Destroys, an episode from the first season of The Human Factor Podcast, I looked at Blockbuster, Kodak, BlackBerry and Sears, organizations that watched the market move and didn’t move with it. The music industry fits the same story from a slightly different angle. The shift starts somewhere outside the established players. Consumers adopt it, first at the edges and then in numbers that are hard to ignore. The industry pushes back and protects what it has, and is eventually pressured into changing anyway. The market drives the outcome. The incumbents get a say in the timing and the terms, but not in the direction.

The Napster story has two lessons for what is happening now. The company that starts a shift is rarely the one that finishes it. Napster didn’t win, Apple and later Spotify did, and whether Muse, Comet or any of today’s agents are still around in five years is anyone’s guess. And the incumbents usually survive, but rearranged. The labels are still here. They just make their money in a way they would never have chosen in 1999.

Being the biggest doesn’t settle it either. The AI tools are far from equal in what they can do, and people treat them a lot like they treat a favorite restaurant, or even a pet that becomes part of the family. They pick a go-to, get attached to it, and some will always chase whatever is newest and getting the most buzz. Having the biggest platform doesn’t guarantee having the assistant people actually want to use, and the first ones out in a market are often not the last ones standing.

Look at what Amazon did in the same month it blocked Muse. It announced a partnership with OpenAI that lets its advertisers run ads inside ChatGPT. Shopify went the other way and opened its stores to Meta’s agent. OpenAI tried letting people buy directly inside ChatGPT, reportedly found that most people used it to research rather than to purchase, and pulled back in March to focus on helping people compare products. Everyone is fighting, testing and negotiating at the same time, which is exactly what the early 2000s looked like in music.

Consumers will find a way, as they always have, and the AI tools will keep finding ways to do what they are asked to do. The friction we see today will eventually give way. What we don’t know yet is how long that takes and what is left standing when it does.

The Next Few Years

Nobody can say with certainty what happens next, and after last week’s issue on the illusion of validity, I’m wary of anyone who claims to. But some of the outlines are already visible.

Expect more friction for a while. More blocking, more lawsuits and more terms of service written to keep agents out, alongside more partnerships that let them in on someone’s terms. Over time, fights like these tend to settle into rules. We are likely to see agents that identify themselves, standards for how they connect to stores and services, and agreements over who gets paid when a machine makes a recommendation or completes a purchase. Some of that work has already started.

Advertising won’t disappear. It will move into the answer. Google is already placing ads inside its AI results, and Schindler said people are clicking on sponsored product listings in AI Mode about as often as the unpaid ones. Amazon is now selling ads inside ChatGPT. The money will follow the attention, as it always has. But putting ads into AI answers is a lot more complicated than putting them on a search results page, and that deserves a closer look.

Organizations will also scramble to understand how these systems decide what to show. That is hard, because the companies building them aren’t saying much, and a good share of what is being sold as “AI optimization” right now is educated guessing under a new name. There is some early research, though. A 2024 study led by researchers at Princeton found that content was more likely to show up in AI answers when it cited credible sources, quoted people with real expertise and included specific evidence. Keyword stuffing, the old trick, did almost nothing, and smaller websites gained the most. Pew found that the sources cited most often in Google’s AI summaries were Wikipedia, YouTube and Reddit, places where people explain and argue about things rather than sell them. If that holds, the organizations that do well in the near term will be the ones genuinely worth quoting, not the ones best at gaming a ranking.

Ads in the Answer Won’t Be That Simple

It is tempting to stop there and assume advertising simply moves into the AI and carries on as before. That is probably too simple an answer.

Picture asking an AI which running shoe is best for someone with flat feet. A sponsored listing for one brand sits at the top. Right below it, the answer recommends a different shoe and summarizes reviews saying the sponsored one wears out within a few months. The advertiser paid to be part of that conversation, and the conversation argued against them. In the old model, the ad and the search results rarely contradicted each other that openly, because the shopper had to do the comparing on their own, across all those tabs. Now the comparison happens in seconds, sitting right next to the ad.

That puts everyone in an awkward spot. If the AI softens its answer to protect the advertiser, people will notice, and they will stop trusting it. Trust is the only reason anyone uses the assistant in the first place. If the AI stays honest, the value of the ad changes, because the advertiser is paying to appear next to a judgment it doesn’t control. Either way, the model that has funded advertising for the last twenty years, paying for position and attention, doesn’t carry over cleanly. Something new will have to take its place. It might mean paying to be included in the comparison rather than placed above it, paying for accurate product information an AI can rely on, or paying only when a purchase actually happens. Nobody knows yet, and the platforms are experimenting in real time.

AI will help build whatever comes next, and it is already doing much of the work. In 2025, Meta said it was aiming to fully automate ad creation and targeting by the end of 2026. An advertiser would submit a product image and a budget, and Meta’s AI would create the images, video and text, find the audience on Facebook and Instagram, recommend how to spend the budget and adjust the ads in real time.

That raises a question very few organizations have asked yet. Who makes the decisions? For decades, the marketer chose the audience and the message, the sales team weighed in on what customers needed to hear, and an agency or a media buyer placed the ads. If a company’s own AI is creating the ads, choosing the audience and moving the budget, and the customer’s AI is deciding which product to recommend, then a growing share of the decisions are being made by machines dealing with other machines on behalf of people. The marketer’s job starts to look less like making those calls and more like setting the goals, the limits and the values the AI works within, and then checking whether what it does still reflects what the organization stands for.

It also raises a harder question about accountability. When an AI-placed ad shows up next to an answer that undercuts it, or reaches an audience nobody would have chosen, who answers for that? The marketer who set the goals, the sales leader who owns the number, the vendor whose system made the call, or nobody at all? Marketing, sales and media buying were all organized around people making choices about other people. Most organizations haven’t decided yet who owns those decisions when much of the work is automated on both sides of the transaction. That may turn out to be one of the larger transformations hiding inside this one, and it reaches into roles, decision rights and the organization chart itself.

The Larger Transformations Behind This One

The near term is the easier part to see. The bigger shifts sit underneath it, and I would suggest they are larger than most organizations are planning for.

The most immediate is who owns the relationship with the customer. For decades, that relationship lived on a company’s website, in its stores, in its catalog and in its customers’ inboxes. If more and more people form their impressions and make their choices through an AI assistant, the assistant ends up in the middle of that relationship. The organization may still make the product and deliver the service, but the first conversation, and maybe the decision itself, happens somewhere else.

Closely tied to that, the customer may increasingly not be a person at all. When an agent is shopping, comparing or booking on someone’s behalf, organizations will need to be understood by the machine as well as by the person it works for. That means clear, accurate and consistent information about what you do and what you offer, the kind an agent can rely on. It is a very different discipline from persuasion, and most marketing functions weren’t built for it.

Then there are the economics that paid for much of the internet. The content and subscription models have been under pressure for years, and many newspapers turned to clickbait simply to survive, chasing the visit because the visit was what paid. A great deal of the free web, the journalism, the reviews, the how-to articles and the forums, was funded the same way. If the visit fades, the money that sustained much of that content comes under even more pressure, and the AI systems that depend on that content for their answers will feel it too. Something will have to replace that arrangement, and we don’t yet know what. That market is going to keep changing, and radically.

It also deepens a shift that was already well underway. Most organizations still communicate one-to-many. They publish, broadcast, send the campaign and wait for people to come. But the way people actually find and judge things has been moving to many-to-many for years, through reviews, forums, social feeds and each other. AI pulls all of those voices together and hands back a single answer. An organization still built around the one-to-many model, still using the megaphone rather than the mirror, is now talking into a conversation that is happening without it.

Underneath all of it is trust. In the old model, people decided whether to trust a brand. In the new one, they may be deciding whether to trust the assistant, while the assistant decides which brands to trust on their behalf. Whose interests the assistant actually serves, the person asking or the company paying for placement- may become one of the defining questions of the next decade. In late 2024, we imagined pieces of this world in AI Mindshifts: The Death of Brands. It read like speculation at the time. Less than two years later, the largest companies in search and retail are fighting over who gets to stand in that spot.

Some organizations will feel all of this sooner than others. The ones likely to hold up best are those an AI treats as the authority on their subject, and those whose members or customers come to them directly because they want to. The organizations in the middle, the ones that depended mostly on being found, will likely feel it first and for the longest. There is a quieter cost as well. Comparing six tabs was tedious, but it was also how people learned to judge for themselves, weighing one review against another and noticing when something didn’t add up. If that goes away, it is the consumer version of the disappearing middle I wrote about in October, where we automate the very work that taught people how to think.

Why Most Organizations Are Slow to Move

If all of this is visible, why are so many organizations still running the same playbook?

Part of it is the investment. When you have spent ten or fifteen years and a great deal of money, building something, walking away from it feels like admitting a loss. Part of it is the measures. Organic rankings and website traffic are still on the monthly dashboard, and as I wrote in September, organizations can add a measure but rarely subtract one. Even the newer measures are shaky. In October, ChatGPT briefly stopped tagging the links it sends people to, and for about a week much of that traffic showed up in website analytics as if visitors had typed the address in themselves. The tag was reportedly restored within days, but most ChatGPT clicks still arrive without the information that would tell a website where they came from. Many organizations can’t reliably see the AI visits they are getting, let alone the ones they aren’t. Part of it is people. There are teams and agencies whose expertise, and whose sense of what they are good at, is tied to the old model. Asking them to accept that what they mastered matters less is the kind of identity renegotiation I wrote about in August.

And part of it is that nobody gets to start over. Most organizations can’t stop everything, go back to ground zero and rebuild how they reach people from scratch. The website still has to work, and customers still search. Meanwhile, the AI tools keep changing month to month, in what they can do and in how many people use them. It is a bit like trying to redesign the train while it is moving and picking up speed.

That is where the difference between change and transformation shows up. Swapping SEO for “GEO,” learning a new set of tactics to get mentioned by an AI instead of ranked by Google, is a change. It is a new way of doing the same job. The music labels made plenty of changes in the early 2000s, and none of them saved the CD. The bigger question is whether the job itself is still the same. If people increasingly come to know you through a conversation you aren’t part of, the question is no longer how to win the click. It is how people come to know you, trust you and choose you at all. Answering that is closer to transformation, and it reaches well beyond the marketing department.

Back to the Running Shoes

So, go back to that pair of running shoes. The person buying them still has the same need, the same budget and the same feet. What changed is the path between the question and the purchase, and for a growing number of people that path no longer runs through your website, your ads or your inbox. It runs through a conversation with a machine that has read a little bit of everything and decides, in a few seconds, who is worth mentioning.

The music industry spent the better part of a decade fighting the path its customers had already chosen. It won in court and lost the market, and by the time it embraced what people wanted, someone else had already built it.

When someone asks an AI about the problem your organization solves, does your name come up, and do you know what it says about you? Are you spending your energy protecting the way people used to find you, or preparing for the way they will find you next? And if the visit never comes, what is your relationship with the people you serve actually built on?

Related Reading

AI Mindshifts: The Death of Brands (Issue 192). December 2024, when we imagined AI assistants doing the searching and marketers being cut out of the picture.

The Readiness Illusion. Why the loudest claims about AI agents tend to run ahead of the evidence, a useful check on any projection, including the ones in this issue.

Who Owns That Number? Why organizations can add a measure and never subtract one, which is why ranking reports outlive the traffic they used to predict.

The Identity Renegotiation. What AI asks people to give up before it gives anything back, including the expertise they spent years building.

The Disappearing Middle (Issue 284). What happens when we automate the work that taught people how to think.

Smart Marketing Is a Mirror, Not a Megaphone (Issue 225). On marketing that reflects the audience rather than broadcasting at it.

How Sure Are You? (Issue 289). Last week, on the illusion of validity and why a story that has held together for years is so hard to question.

On The Human Factor Podcast

Who Leads When AI Thinks?, Season 3 Episode 039. With Adrian Wolfberg, on judgment, ambiguity and who decides when AI is part of the decision.

The Drift That Destroys, Season 1 Episode 011. Blockbuster, Kodak, BlackBerry and Sears, and why successful organizations drift away from where the market is heading through thousands of small, reasonable decisions.

Research Sources

1. Pew Research Center (2025). Google Users Are Less Likely to Click on Links When an AI Summary Appears in the Results. July 22, 2025. Source of the 900-adult browsing study, the 8 percent and 15 percent click rates, the 1 percent click rate on summary links, the higher session-ending rate, and Wikipedia, YouTube and Reddit as the most cited sources.

2. Gartner (2024). Gartner Predicts Search Engine Volume Will Drop 25% by 2026, Due to AI Chatbots and Other Virtual Agents. Press release, February 19, 2024.

3. Alphabet Inc. (2026). Alphabet Announces Third Quarter 2026 Results. October 28, 2026. Source of Google Search and other revenue ($63.35 billion in Q3 2026 against $56.57 billion in Q3 2025) and Sundar Pichai’s statement on query growth.

4. Alphabet Q3 2026: Search Ad Growth Slowed, Queries Did Not. Security Boulevard, October 2026. Source of the second-quarter growth comparison and of Philipp Schindler’s earnings call remarks on currency, AI Mode monetization and sponsored listings in AI Mode.

5. Amazon Blocks Meta’s Muse and Accuses Perplexity of Misleading a Court. The Next Web, September 2026. Source of the Muse launch and block dates, Amazon’s objections, Meta’s response, and the Perplexity injunction and Ninth Circuit reversal.

6. Amazon’s $68 Billion Reason to Block Meta’s Muse. Forbes, September 23, 2026. Source of the advertising revenue figure and Shopify’s decision to open its storefronts to Muse.

7. OpenAI’s Plans to Make ChatGPT More Like Amazon Aren’t Going So Well. TechCrunch, March 24, 2026. Source of the Instant Checkout pullback and OpenAI’s shift to product discovery.

8. Amazon Ads (2026). Amazon Ads Announces Partnership with OpenAI to Support Advertising in ChatGPT. September 10, 2026.

9. Aggarwal, Pranjal, Murahari, Vishvak, Rajpurohit, Tanmay, Kalyan, Ashwin, Narasimhan, Karthik, and Deshpande, Ameet (2024). GEO: Generative Engine Optimization. Proceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, August 2024. Source of the finding that citing sources, adding quotations and adding statistics improved visibility in AI answers, the weak result for keyword stuffing, and the larger gains for lower-ranked sites.

10. Napster Released. EBSCO Research Starters. Source of Napster’s June 1999 launch, the December 1999 RIAA lawsuit, the Metallica suit, the 2001 court order and the July 2001 shutdown.

11. RIAA Begins Suing Individual Sharers of Copyrighted MP3 Files. History.com. Source of the September 8, 2003 announcement and the 261 individuals named in the first round of lawsuits.

12. Apple (2003). Apple Launches the iTunes Music Store. Press release, April 28, 2003. Source of the 99 cents per song price and the participation of BMG, EMI, Sony, Universal and Warner.

13. Recording Industry Association of America (2017). 2016 Year-End Music Industry Revenue Report. Source of streaming’s growth from 9 percent of revenue in 2011 to 51 percent in 2016, and of 2016 revenue standing at about half of 1999 levels.

14. Meta Aims to Fully Automate Advertising with AI by 2026. exchange4media, June 3, 2025, reporting Reuters and The Wall Street Journal. Source of Meta’s goal to fully automate ad creation and targeting by the end of 2026 from a product image and a budget.

15. ChatGPT UTM Tag Loss Was a Bug, Says Adobe’s Huesler; Fix Rolling Out. Relevant Audience, October 2026. Source of the October 8 to 15 loss of ChatGPT link tagging, the fix, and the continued absence of referrer data on most ChatGPT clicks, as reported by Adobe’s Cedric Huesler. OpenAI has not commented publicly.