Google didn't lose money. It lost the click.

A post doing the rounds says Google forced AI summaries on everyone, people stopped clicking and now it's losing billions in ad revenue. It's half right. The wrong half matters, because if you get the diagnosis wrong you'll get the response wrong too.


What actually happened

Alphabet reported its July to September results on 28 October. Revenue was up 25% to $127.6bn. Search revenue was up 12% to $63.35bn. That isn't a company losing money.

Profit did fall by about a quarter, to $26.2bn. Most of that came from an $8.7bn loss on investments it holds in other companies. It had nothing to do with search.

So where's the story? Investors expected around $64.85bn from search and got $63.35bn. Compared with the previous three months search revenue barely moved: up $79m. Over the same period Google says searches grew faster than at any point in five years.

More searches. Flat money. On the investor call Google said its own AI Mode was showing fewer ads. The next day Alphabet's shares had their worst day on record and around half a trillion dollars came off the company's value. Only Nvidia has lost more in a single day.

Was it the US midterms? No. The fall came on 29 October, five days before the vote. Tech shares were nervous in the run-up, partly over whether a new Congress would be tougher on AI, but the trigger was the search figures. When the results came in, Alphabet's shares went up, not down, rising 3 to 4% the day after the election as big tech rallied.

Google hasn't lost billions. It has found out that an answer earns it less than a page of links. That was always the risk and it was always going to land eventually.

Where the ads go next

The fix is obvious: put the ads inside the answer. Google is already doing it. On 28 October it switched on ads inside AI Mode answers in ten more countries. The UK is due before the end of the year. The EU isn't, for now.

Advertisers can't target AI answers on their own or opt out of them. Existing Search, Shopping and Performance Max campaigns are eligible automatically and the results are folded into ordinary campaign reports. If you run Google Ads you may already be in there and have no easy way to tell.

The label problem

For now every one of those ads carries a "Sponsored" label. The worry isn't that the label disappears. UK and EU consumer law already bans paid content dressed up as editorial, so it won't. The worry is that it stops meaning anything.

People already struggle. When Ofcom showed UK adults a page of Google results in 2015, with the top three ads in an orange box marked "Ad", half of them didn't realise they were ads. That was a separate block at the top of a list. Now picture the same ad sitting in the middle of a friendly paragraph that's answering your question.

There's an old line in this industry that most people don't trust paid results and click on organic ones instead. It was never a precise figure. It was a nine-out-of-ten-cats sort of claim. But it held because people could see where one ended and the other began. Inside an AI answer that line gets very thin: small label, same voice, same box.

That's good for Google's revenue and bad for anyone relying on the old split.

Will UK search go the EU's way?

Search results already differ either side of the Channel. EU users don't get ads in AI Mode yet. UK users will by the end of the year.

Closer ties with the EU won't change that soon. The UK and EU meet for a reset summit on 20 November, but the reported agenda is youth mobility, food standards and carbon markets, not digital rules. The government says it's staying outside the single market for now.

The bigger change for UK search is home-grown. In June the Competition and Markets Authority used its new powers to impose rules on Google. Publishers must be able to opt out of AI Overviews without dropping out of search. Their content must be properly attributed with clear links. Google must show the regulator that it ranks results fairly, including in AI Overviews and AI Mode. It has nine months to comply, which takes us to around March 2027.

So if UK results change next year, it'll be because of a British regulator, not Brussels. Worth watching: an AI Overview that has to link clearly to its sources could send UK sites more traffic than the same answer elsewhere.

Optimising for AI answers alone is pointless

"Optimise for AI" is the phrase of the year. Most of it misses the point. If the answer gives people everything they need nobody clicks. Being cited in an answer that sends no traffic is a nice screenshot, not a business.

So the work has to be paid and organic together. SEO earns the place in the answer. PPC buys the slot inside it. Until now PPC teams had no reason to care about featured snippets and rich results because they couldn't bid on them. Now they can, and they'll have to, because that's where the clicks that are left will be.

If SEO and PPC still sit in separate teams with separate reports, they'll spend a lot of money competing with each other for the same few clicks.

Stop chasing the long tail

The long tail and the "low hanging fruit" are the first things AI answers swallow. They're the simple, specific questions an answer box handles best. Putting your effort there now is building on sand.

The searches that still click are the ones where people already know who you are: your name, your product, your shop. That means building a brand. It's slow, it doesn't fit neatly into a monthly report and it's the only plan that will still work in five years.

Building a brand, from an SEO point of view

Brand sounds like the marketing team's job. A lot of it is SEO work:

AI makes life easier, if you know what you're doing

You'll hear that AI turns a day's work into five minutes. Sometimes it does. But the five minutes only works if you already know what good looks like. AI speeds up someone who knows the job. It doesn't replace knowing the job. An agency selling five-minute work without the expertise behind it is selling a faster way to get things wrong.

Ecommerce went the same way. It didn't win because it was clever. It won because buying things got easier: no queue, no opening hours, no trip into town.

AI shopping agents are the next step. Someone asks an assistant to find and buy something and the agent does the legwork. The sites that make that easy will win. The ones with a convoluted checkout, thin product data and six pop-ups will be skipped, and the customer will never know they existed.

The tools are already here. An agents.md file tells AI agents how to search your catalogue, build a basket and start a checkout. Shopify now generates one for every store by default. This site has one of its own. Behind it sit checkout standards such as the Universal Commerce Protocol from Google and Shopify and the Agentic Commerce Protocol from OpenAI and Stripe. Most sites haven't looked at any of it.

Loyalty is earned by being easy

People aren't loyal to brands that make life difficult. They put up with them until something easier comes along. Building a brand and making buying easy aren't separate jobs. The brand is why someone asks for you by name. Ease is why they come back.

So are agents.md and conversion rates the new brand? No. They're part of it, and a bigger part than they used to be, because there are now two audiences.

The first is still a person. A brand puts you in their head, so they say "find me running shoes from you" and not just "find me running shoes". No file or checkout tweak does that.

The second is the agent. It has no feelings about your logo. It chooses on whatever it can read and act on: clear product data, stock, price, delivery, returns and a checkout it can complete. To an agent, how easy you are to buy from is the brand, because it's all it can see.

Conversion rate used to measure how well a page turned visitors into buyers. It now needs a wider definition: how easy you are to buy from, by a person or an agent. Some of those sales won't show up as visits to your site at all, because an agent can complete a checkout inside Google's AI Mode. If you only watch your own analytics, you could miss the shift completely.

Google's quarter isn't a crash. It's a sign of where the clicks are going. Plan for fewer of them, make the ones you get count and make your site the easiest one to buy from.

Sources: Alphabet Q3 2026 results, PPC Land on search revenue, NBC on the share price fall, Moneywise on the search ad miss, TIKR on the post-midterm rally, PPC Land on AI Mode ads, Econsultancy on Ofcom's search ads finding, Mondaq on the CMA's rules for Google, Shopify on agents.md