Q4 is here and BFCM won’t wait. On Nov 18 (1PM EST / 7PM CET), Pixii AI’s 15X5 Amazon Hacks Webinar puts 18 working Amazon operators in one room for 90 minutes of rapid-fire, ready-to-use tactics. There are no 45-minute decks and no sales pitches, just plays you can put to work the same week.

You’ll get PPC lessons from 100+ account audits, the profit killers hiding in your bestsellers, and AI workflows that audit an entire account in 15 minutes. You’ll also see insights from 100K top-performing listings, how to get found on Google, LLMs and Alexa, and BFCM velocity and repeat-purchase plays.

If you sell on Amazon or run a brand there, this is the fastest way to take what’s working right now. Seats are free, so grab yours before BFCM hits.

For twenty years, paid search was the one line in the media plan nobody had to defend. Social got interrogated every quarter. Connected TV got the raised eyebrow. Influencer budgets got the full Spanish Inquisition. Search got renewed, usually without anyone looking up from their laptop.

Alphabet's third-quarter results have made that slightly awkward.

Google just reported its fastest query growth in five years. Search revenue, meanwhile, grew 12% to around $63 billion, down from 19% in the first quarter and 17% in the second. It's a shop reporting record footfall while the till stays suspiciously calm, because everyone's getting their questions answered at the door.

Google offered three explanations, which is two more than anyone offers when things are going well.

The first was currency, the corporate equivalent of "the dog ate my homework", except occasionally the dog genuinely did. The second was weak advertiser categories, including Asian retailers hit by the EU scrapping its customs exemption for low-value parcels (closing a loophole, it turns out, has consequences). The third was AI Mode, Google's conversational search, where queries are booming but each one earns less.

Only one of those three is structural, and it's the one Google built itself. This is the first time the company has said out loud that its own AI product is nibbling at its own search revenue, which is a bit like a bakery admitting the new free-samples counter is affecting bread sales.

Before anyone drafts Google's obituary (again): it remains spectacularly profitable, and retail was still its biggest search ad growth driver. Nobody needs to pass round a collection tin. The auction is just changing shape underneath us.

The classic results page makes money because shoppers have to do homework. They scan, compare, refine, search again, and every one of those moments of mild confusion is a sellable impression. Google has spent two decades monetising our indecision, and in fairness, we've given it plenty to work with.

AI answers do the homework for us. Google says AI Mode queries are on average three times longer than normal searches. The shopper who used to run six short searches for trail running shoes now asks one long, slightly over-specific question and gets one considered answer. Brilliant for the shopper. Less brilliant for an auction priced on how many times they had to ask.

AI Mode also carries far fewer ads. Two pressures, then: fewer clicks to fight over on the classic page, sparse inventory inside the shiny new one. The second is Google's problem. The first is ours, and it arrives via the invoice.

Most Google campaigns now run on automated bidding, where Google's systems adjust bids to hit your targets. When competition for the remaining clicks heats up, those systems simply pay more to win them (Smart Bidding is very smart; whose interests it's being smart about is a question for another issue).

One large agency reported branded cost per click up about 5% in Q3. One agency's data, and clicks grew too, so this isn't purely scarcity. But it matches what buyers describe, and Q4 is always the priciest quarter.

The maths is unglamorous. If CPCs rise 5% while conversion rate and order value hold steady, ROAS falls by roughly 4.8%. A brand earning £4 per £1 spent now earns about £3.81. Same product. Same listing. Same checkout you've rebuilt twice. The customer just got more expensive, and after fees and fulfilment, those 19 pence can be the difference between a profitable campaign and one quietly setting fire to your margin.

Google's answer is more ads inside the AI surfaces. Ads in AI Mode are expanding to ten more markets, with Brazil, Mexico and the UK scheduled for this quarter. UK sellers: this is no longer an American problem you get to read about with mild detachment.

In the US, Google has also launched multiturn ads, which appear in the follow-up questions of a conversation. Picture a shopper asking which trail shoes suit wide feet, then which cost under $120. That second question is the precise moment someone goes from browsing to shortlisting. It's commercial gold. Google says click-through rates on sponsored listings in AI Mode are comparable to organic ones, based on its own US data, over two whole weeks, a sample size I'll politely describe as confident.

Two catches. First, control. Advertisers reach most of this inventory through AI Max and Performance Max rather than by choosing keywords, so brands get less say over which conversations they turn up in. Unless Google reports AI Mode separately, we'll be taking its value on trust. A bold ask, from the people who brought you Performance Max transparency.

Second, the user. Around 70% of US and UK consumers who regularly use AI answer engines say they trust what they're told. That trust is what's driving the query growth. Google needs enough ads in AI Mode to recover the revenue and few enough that nobody starts wondering whether the answer was sponsored. It's a tightrope, and advertisers are paying for the rope.

Across recent accounts from eighteen media buyers, typically 10–20% of Google search budgets are under review (some agencies say 30%). Search is getting the scrutiny usually reserved for the experimental line item nobody wants to own.

The method is shifting too. Agencies running media mix modelling and geographic holdout tests (switch spend off in some regions, leave it on in others, compare) report that video and social contribute more to sales than attribution suggested, and search less.

Here's the uncomfortable bit. Platform ROAS credits an ad with every sale after a click. Incremental ROAS asks how many of those sales would have happened anyway. A shopper who types your brand name into Google was going to buy regardless; the click takes the credit, the persuasion happened elsewhere. Branded search has always been the colleague who presents the deck somebody else built.

AI assistants widen that gap. A shopper compares options in ChatGPT, settles on a brand, then Googles it to buy. Google gets credited with a decision made inside a competitor's product. The dashboard stays green. It will, in fact, be the very last thing to notice.

Where the reviewed money goes is messier than the headlines. One agency saw US clients' Facebook spend rise 16%. Another says most diversification stays within Google, shuffling into AI Max, Performance Max, YouTube and Demand Gen. Search may be losing its hall pass without Google losing the budget. Nicely played, Mountain View.

The obvious alternative is ChatGPT: 1.2 billion weekly users and search-like query volume around a fifth of Google's (OpenAI's own figures; it hasn't said how it counts a query). Brands advertising there grew nearly sixfold between June and August, and the tooling has grown up just as fast.

Performance hasn't kept up. One working paper found paid ChatGPT visits generated $0.89 of revenue per session versus $2.45 for paid Google Search. It's a single study, still under review (yes, I'm citing a study and doubting it in the same breath; it's called balance). But at a ROAS target of 3, that's the difference between affording an $0.82 click and a $0.30 one. One analyst model had assumed ChatGPT CPCs near $5, and has since suggested that might be a touch optimistic. Somewhere, a spreadsheet is having a quiet word with itself.

Now the strange part. Shoppers arriving via links inside organic AI answers converted 90% better than other traffic in September. So the shopper an AI recommends you to is gold, and the shopper an AI is paid to show you is... not. Same assistant. Wildly different outcome.

Timing might explain it. An organic referral arrives at the end of a conversation, once the comparing is done. An ad may arrive earlier, or land as an interruption in a chat the user thought was impartial. It's the difference between a friend recommending a restaurant and the owner pulling up a chair at your table to talk you through the specials. Better formats might close that gap. Or the gap might be telling us something more permanent.

None of this says abandon paid search. Google is still the most dependable way to reach people who already know what they want. It just shouldn't be exempt from the standard of evidence every other channel has to meet.

Amazon sellers will recognise the plot. Sponsored Products now extends into Rufus answers, reported by several accounts inside standard Sponsored Products metrics rather than broken out. Same film, different cinema. Every discovery surface is learning to sell ads inside a conversation, and in every case the measurement is turning up after the inventory. Fashionably late, as ever.

Can Google raise ad density in AI Mode without spending the trust that's driving the usage? Is ChatGPT's conversion gap an immature ad product or a permanent feature of ads inside a chat? Will measurement grow up fast enough for budgets to move on evidence rather than gut feel?

And underneath sits one we've never really had to ask. For two decades, a search ad was a bet placed at the exact moment a shopper said what they wanted. If that moment now happens inside an assistant, before any results page loads, what exactly are we paying for when we pay for the click?

Jo

P.S. If your platform ROAS looks fantastic right now, that's either great news or precisely the problem. A January holdout will tell you which.

P.P.S. Google once tested replacing its "I'm Feeling Lucky" button with AI Mode. At the time it looked like a product experiment. In hindsight, it reads more like advice for advertisers.

Put your brand in front of an engaged audience of 70,000+ subscribers, with 75% based in the US.

Interested in sponsoring our newsletter? Get our media kit & rates.

🎥 Pictory GPT - Effortlessly create captivating videos

🌐Dora AI - Create a fully editable website from just one line of text with AI.

🎨 ReScroll - The AI-powered visual board and knowledge base.

📂 Guidde -Magically create video documentation with AI.

ABOUT THE WRITER:

Jo Lambadjieva is an entrepreneur and AI expert in the e-commerce industry. She is the founder and CEO of Amazing Wave, an agency specializing in AI-driven solutions for e-commerce businesses. With over 13 years of experience in digital marketing, agency work, and e-commerce, Joanna has established herself as a thought leader in integrating AI technologies for business growth.

For Team and Agency AI training book an intro call here.