AI shopping picks a different seller. Agents make free returns expensive.

In the fifteenth episode of Assistant to the CMO, Lexie Meskouris speaks with Will Nash and Harjot Singh about what happens when an AI shopping shelf quietly picks a different seller for a product you make.

Productrise ran the same shopping queries through Google's AI Mode and through ordinary search, in both the US and the UK. The first price shown in AI Mode averaged 21.6% higher, and where prices differed a different seller sat behind the listing about half the time. Will thinks a snapshot of a surface Google keeps changing is worth very little. Harjot stakes the opposite, that the shelf favours whoever looks like the place to buy, and that the premium attached to that is permanent.

The second half turns to Meta's new shopping agent and what Will thinks it breaks first. Generous returns policies are often priced on customers not bothering to claim, and an agent removes that assumption for free. The advice is small: check who the shelf says is selling your product, and read your refund terms the way a bot would.

Highlights of this Episode Include:

• The Shelf Chooses the Seller: Where prices differed, a different seller sat behind the AI Mode listing 49.6% of the time.
• The Headline Measured a Model That Moved: Will recalls AI Mode changing soon after publication, leaving the price number close to worthless.
• Canonical Beats Cheapest: Harjot stakes that Google optimises for the best place to buy, and that a premium for the canonical seller is permanent.
• Google's Defence Is a Non-Answer: The Shopping Graph is a data store, so it says what is available and nothing about what gets chosen.
• The Brand May Not Be the Shop: Harjot concedes the canonical place to buy is often a retailer with more evidence behind it than the manufacturer.
• Free Returns Were Priced on Apathy: Meta's shopping agent is live, and Will argues offers that assume nobody chases a refund are exposed.
• Two Things to Do This Week: Check the lead seller and price for your three best sellers in AI Mode, and read your refund terms as a bot would.

Listen and subscribe: https://AssistantToTheCMO.show

Learn more about Aloudable: https://aloudable.com/

Book a demo: https://aloudable.com/get-a-demo

New episodes of Assistant to the CMO land every Thursday.

ChatGPT ads promise not to touch the answer. Google expands AI Overviews and gives no numbers.

In the fourteenth episode of Assistant to the CMO, Lexie Meskouris speaks with Will Nash and Harjot Singh about whether anybody outside OpenAI can ever check the promise that ads do not influence the answers ChatGPT gives.

OpenAI says ChatGPT Ads reached a one billion dollar annualised run rate under 200 days after its first ad, with self-serve buying now open across Europe, India, the Middle East and North Africa. Both call that signal. The question underneath is whether OpenAI's promise that ads do not influence answers can be checked from outside. Harjot Singh calls it policy rather than a technical guarantee, and Will Nash argues a regulator can only police what leaves a trace.

The second half is Google auto-expanding AI Overviews, with no numbers given. Harjot's read is that the old length limit was economic and layout rather than retrieval. Will's worry runs further out, because removing how publishers get paid makes it harder for the models to stay current. The Monday move is to check whether your buyer sits on a ChatGPT tier that can see an ad.

Highlights of this Episode Include:

• Somebody Else's Customers: Ads run only for logged-in adults on Free and Go, so check which seat your buyer sits in before spending.
• It's Always Policy: Harjot on why OpenAI's promise that ads do not touch the answer has no mechanism under it.
• The Test That Can Never Close: Harjot's probe ad would only ever give suspicion, because targeting is the same signal as relevance.
• Plausible Deniability Is the Shape of It: Will on why a regulator can only police what leaves a trace, and a model's framing leaves none.
• Policing Something Nobody Can See: Harjot's earlier targets policed a label anyone could check, which is why this one needs a regulator.
• The Short Answer Was a Layout Decision: Harjot argues the old length limit came from economics and screen space rather than retrieval.
• Where the Answers Come From in Five Years: Will on how removing the way publishers get paid makes it harder for the models to stay current.

Listen and subscribe: https://AssistantToTheCMO.show

Learn more about Aloudable: https://aloudable.com/

Book a demo: https://aloudable.com/get-a-demo

New episodes of Assistant to the CMO land every Thursday.

Retailers can't out-build the shopper's AI. Amazon lost in court trying to block it.

In the thirteenth episode of Assistant to the CMO, Lexie Meskouris speaks with Will Nash and Harjot Singh about what a retailer can do when the shopper arrives holding their own AI, and whether either of the two obvious answers still works.

Narvar's US-only 2026 holiday report found 65 percent of shoppers plan to use AI for part of their Christmas shopping, and 8 percent of retail decision-makers feel very confident using it. Will Nash would track the 43 percent doing gift discovery instead, and adds a third source for a model's opinion of your brand: your live product catalogue. Harjot Singh argues a retailer cannot out-build the shopper's own agent, which knows the customer when the retailer's site starts from nothing.

The second half is the Ninth Circuit vacating Amazon's injunction against Perplexity's Comet, which turned on where a screenshot gets processed. Will has no view on the case. Harjot argues nobody designed that architecture for a court, GPU costs did, and that blocking agents is futile. The Monday move is smaller: put your returns and shipping terms on the product page as plain text.

Highlights of this Episode Include:

• The Number That Costs You Money: Will skips the 65 percent headline for the 43 percent of US shoppers using AI for gift discovery.
• It Knows You, Your Website Doesn't: Harjot's case that a retailer cannot out-build an assistant the shopper already carries.
• A Third Layer, and You Maintain It: Will adds the live product catalogue to memory and search as a source of what a model thinks of you.
• It Cannot See Your Photographs: Harjot on why shopping indexers mostly read text, because classifying every image costs too much.
• Amazon Lost on GPU Economics: The Ninth Circuit ruling turned on where Comet's screenshot is processed, which Harjot says hardware decided.
• Nobody Is Selling You the Future: Harjot on vendors pitching agent-ready commerce when the thing itself is still an experiment.
• Put the Returns Terms in Words: Will's move for when price is level, and Harjot's test is whether the page still reads with images off.

Listen and subscribe: https://AssistantToTheCMO.show

Learn more about Aloudable: https://aloudable.com/

Book a demo: https://aloudable.com/get-a-demo

New episodes of Assistant to the CMO land every Thursday.

In the twelfth episode of Assistant to the CMO, Lexie Meskouris speaks with Will Nash and Harjot Singh about what a company can measure and own, and whether either one tells you what it can actually do.

OpenAI published two sets of numbers this month showing its heaviest enterprise users pulling away from everyone else. Will Nash reads it as a cost metric dressed as maturity, since OpenAI's "frontier firms" are simply its biggest spenders. Harjot Singh reads it as a mirror of the power-user distribution that always existed on individual plans. They agree the number is weak, then split sharply on what should replace it.

The second half turns to Omnicom, which moved at least 468 people who built its Omni AI platform to a contractor while still owning the platform, the data and the clients. Will argues that owning data you cannot change is worth little. Harjot argues 468 is not a big number and the real failure sits with leadership. Both expect the product to drift from the company selling it, and that is the question to ask at renewal.

Highlights of this Episode Include:

• Frontier Firm Means Big Customer: OpenAI's top usage decile measures spend, and Will says a cost line is being presented as innovation.
• Tokens Are Not Usage: A legal team running long contracts burns tokens a hundred short prompts never will, so the growth curve misleads.
• Enterprises Are Mini Networks: Harjot explains how one person's AI habit spreads to co-workers and compounds into exponential usage.
• Marketing Ranks Fourth For a Reason: Will argues its repeatable work was automated years ago by software that does not bill in tokens.
• A Privilege You Cannot Abuse: Will's line for Omnicom keeping data it can no longer maintain, improve or reshape without calling a supplier.
• 468 Is Small, or It Is the Tell: Harjot calls it routine Pareto maths, Will calls it too large to be only the engineering team.
• One Hour and a Blank Page: Harjot's Monday move is to map a workflow, then ask whether you would build it that way starting today.

Listen and subscribe: https://AssistantToTheCMO.show

Learn more about Aloudable: https://aloudable.com/

Book a demo: https://aloudable.com/get-a-demo

New episodes of Assistant to the CMO land every Thursday.

Google Ads forces AI Max in September. Verification becomes the agency pitch.

In the eleventh episode of Assistant to the CMO, Lexie Meskouris speaks with Will Nash and Harjot Singh about what happens to a marketing budget when the platform spending it stops showing you what you bought.

From 1 September, Google automatically moves Search campaigns using automatically created assets or campaign-level broad match into AI Max, and since 3 August you cannot create either setting anywhere. Will opens by calling it a headline rebrand, then changes his mind on air: the keyword-level report goes, and with it your ability to say what you bid on. Harjot reads the same change as the next rung on a ladder running from URLs to keywords to intent.

The second half turns to WPP, which cut 1,267 roles in six months while pitching itself as a platform. Will thinks the agency of the future sells verification. Harjot calls it a feature that ends up inside somebody else's product. Neither settles it, and the advice they agree on is small: record your baseline, migrate before Google does it for you, and change nothing else while it happens.

Highlights of this Episode Include:

• Creation Closed on 3 August: You can no longer make campaign-level broad match or legacy automatically created assets anywhere, including the API.
• Will Talks Himself Off His Own Verdict: He opens on "a headline rebrand" and ends up arguing you will not know what you are bidding on.
• The Keyword Report Is the Real Loss: Reporting cost per click at keyword level goes away, because queries are mixed together after the migration.
• A Results-Only Relationship: Will's line for what advertisers are left with is that Google defines the product it sold you after it has sold it.
• Cheaper Clicks Are Not Good News: Google can infer commercial intent from searches that touched no keyword, and gets paid on every extra click.
• Verification Splits the Room: Will proposes semantic clustering as a real service line. Harjot says it is a feature and therefore not a business.
• Migrate Before You Are Migrated: Take your baseline now, switch on your own terms before the deadline, and run nothing else across the window.

Listen and subscribe: https://AssistantToTheCMO.show
Learn more about Aloudable: https://aloudable.com/

Book a demo: https://aloudable.com/get-a-demo

New episodes of Assistant to the CMO land every Thursday.

In the tenth episode of *Assistant to the CMO*, Lexie Meskouris speaks with Will Nash and Harjot Singh about whether AI discovery is worth real money yet, and whether you can buy your way into the evidence an AI reads.

Shopify reported that AI-driven traffic and orders to its merchants both tripled year on year, and said AI is a complement to search rather than a substitute. Both men accept the growth and distrust different halves of the report. Will questions the number, a slowdown on Shopify's own earlier figure with no denominator. Harjot questions the sentence, calling the complement framing lazy or disingenuous.

That splits them on money. Will would not move budget yet, since the structures worth optimising for are not settled and big spend now gets wasted. Harjot would move at least fifteen percent of the SEO budget into experiments now. The second half turns to agencies selling Reddit seeding, which Will refuses outright and Harjot would only build in-house. The practical move is smaller than both: write the questions you want to be the answer to, then find who owns the facts you are missing.

Highlights of this Episode Include:

• Tripled, But Decelerating: Will recalls Shopify citing near thirteen times growth in Q1, so tripling is a slowdown, and no denominator is given.
• Complement, or Just Reassurance: Harjot calls the complement-not-substitute line lazy or disingenuous, and aimed at stopping merchants panicking.
• Three Layers of Retrieval: Live search, a crawler-built vector index, and what is baked into the weights, each with very different lag.
• A Model Doesn't Care About Your Brand: Harjot argues live search flattens brand equity, so a day-one brand can compete with an incumbent.
• Fifteen Percent, or Wait: Harjot would move fifteen percent of SEO budget now. Will backs a test but not a yearly escalation.
• The Reddit Deck Nobody Would Sign: Will calls it fake reviews under UK rules, and Harjot gives the tactic six to eight months at best.
• Ten Questions and a Name: Write the questions you want to be the answer to, run them, then name who owns each fact you're missing.

Listen and subscribe: https://AssistantToTheCMO.show

Learn more about Aloudable: https://aloudable.com/

Book a demo: https://aloudable.com/get-a-demo

New episodes of *Assistant to the CMO* land every Thursday.

AI creative settles into a house style. Brand advertising pays for it first.

In the ninth episode of Assistant to the CMO, Lexie Meskouris speaks with Will Nash and Harjot Singh about why AI-generated creative is starting to look the same everywhere, and whether getting out of it is a quick fix or a strategic choice.

Both hosts call the problem real, then split on the cause. Will Nash sees convergence built into the models, from the small pool everything is built on to a reward process that selects whatever the fewest people object to. That, he argues, forces a choice between chasing scale and protecting a design language. Harjot Singh says it comes from taking the output as it arrives, and that the cure costs an afternoon.

The second half turns to New York's new law requiring a label on AI-generated performers. Will treats it as a compliance matter to hand over and wait out. Harjot calls it vanity regulation aimed at the wrong problem, and they split on whether a label repels audiences or attracts them. Neither would build a brand around a synthetic face yet, and both end up at the same practical move, which is to give the model something only your company has.

Highlights of this Episode Include:

• Default Use Creates the Sameness: Harjot argues most people take the first good output and stop pushing, so everyone ends up in the same place.
• The Speed Trap: Work that took four days now takes thirty minutes, and that feels like an achievement until everyone else clears the same bar.
• Why the Models Converge: Will points to a small pool of image models and a reward process that selects the option fewest people object to.
• They Disagree on the Mechanism: Harjot rejects Will's model-scarcity account, blaming a loop where models train on what other models wrote.
• Where Sameness Actually Costs You: Will argues convergent work barely hurts click-driven marketing, but gets expensive in brand and reach.
• New York Labels Synthetic Performers: Will hands it to compliance and waits. Harjot calls it vanity regulation aimed at the wrong problem.
• The Test for Your Own Prompt: If someone at another company could paste your prompt and it still made sense, there is not enough of you in it.

Listen and subscribe: https://AssistantToTheCMO.show

Learn more about Aloudable: https://aloudable.com/

Book a demo: https://aloudable.com/get-a-demo

New episodes of Assistant to the CMO land every Thursday.

In the eighth episode of Assistant to the CMO, Lexie Meskouris speaks with Will Nash and Harjot Singh about what happens when the AI answer becomes both an advertising surface and a measured one.

The episode opens on ads moving inside AI answers, now alongside roughly a quarter of Google's AI Overviews. Both agree it is a real shift that dwarfs the ChatGPT ad pilot at Google's scale. They split on who benefits. Harjot sees cheap, underpriced slots and early-mover advantage, because the conversation itself is the sharpest targeting signal advertisers have ever had. Will is warier, arguing the brands already baked into the models when they trained hold the edge.

The second half turns to Google's new Search Console reports and its crackdown on gaming AI citations. Will sees GEO becoming a real, measurable discipline. Harjot calls the early tools fake comfort and warns that an impression only matters when it moves real traffic. The practical takeaway: run one small AI-ad experiment now to learn the channel, and treat the new data as a signal, not a target.

Highlights of this Episode Include:

• The Ad Moves Inside the Answer: Ads now sit inside about a quarter of Google's AI Overviews, a far bigger shift than the ChatGPT pilot.
• The Conversation Is the Target: The chat is the richest targeting signal advertisers have had, which is also what makes it feel intrusive.
• Cheap Slots, So Experiment Now: With the slots underpriced, Harjot's advice is to run one small AI-ad campaign to learn the channel, not to win.
• Incumbents May Already Hold the Edge: Will argues brands prolific when the models trained are baked in, while others effectively pay to appear.
• Google Starts Scoring the Answer: New Search Console reports show how often you appear in AI answers, and gaming citations now risks a penalty.
• Split on Whether GEO Is Real: Will sees an auditable discipline emerging; Harjot calls the early measurement tools fake comfort.
• Grill the Vendor, Correlate the Data: Press GEO sellers on structure versus citations, and trust an impression only when traffic moves with it.

Listen and subscribe: https://AssistantToTheCMO.show

Learn more about Aloudable: https://aloudable.com/

Book a demo: https://aloudable.com/get-a-demo

New episodes of Assistant to the CMO land every Thursday.