This week brings a clutch of developments that directly affect how we plan paid media, organic visibility, and AI-readiness strategies for our clients. ChatGPT is maturing its ad platform at speed, Google is auto-upgrading Search campaigns to AI Max, and a sharp piece of analysis has exposed why some popular GEO tactics rest on shaky evidence. Here is what matters and what we are doing about it.

Key Takeaways

• ChatGPT Ads now supports oCPC bidding, product carousels, and expanded measurement — making it a viable paid channel for e-commerce brands.
• Google will auto-upgrade eligible Search campaigns with automatically created assets to AI Max from 1 September 2025.
• The llms.txt hype has been dismantled by a simple cats.txt experiment, raising serious questions about unproven GEO tactics.
• Google DeepMind’s WeatherNext 2 demonstrates how rapidly AI model capability is advancing, with implications for every data-driven discipline.
• Emerging product categories can be identified early in search data — if you apply the right analytical framework.

ChatGPT Ads Gets Serious With oCPC, Carousels, and Measurement

OpenAI is no longer experimenting. ChatGPT Ads has rolled out optimised cost-per-click campaigns, advertiser attribution measurement, and product carousel formats, giving performance marketers the tools they actually need to justify spend. For our e-commerce clients, the product carousel format is the headline feature: it brings catalogue-level merchandising into conversational AI for the first time.

We are already testing feed-based campaigns within ChatGPT for select retail accounts. The oCPC model means we can optimise toward conversions rather than raw clicks — a prerequisite before we allocate meaningful budget. If your agency is still treating ChatGPT ads as experimental, the window for early-mover advantage is closing fast.

Google’s AI Max Auto-Upgrade: What You Need to Check Before September

Google has confirmed it will auto-upgrade eligible Search campaigns using automatically created assets to AI Max from 1 September. AI Max expands search term matching and enables real-time text customisation powered by Google’s generative AI.

This is not optional. If you have campaigns with automatically created assets enabled, Google will flip the switch unless you opt out. Our team is auditing every client account this month to:

• Identify campaigns that qualify for the upgrade.
• Review search term reports and negative keyword lists to prevent irrelevant matching.
• Decide on a campaign-by-campaign basis whether the broader matching aligns with the client’s goals.

Broad AI-driven matching can unlock incremental volume, but it can also burn budget on low-intent queries. Preparation now saves waste later.

The Cats.txt Experiment Exposes Weak GEO Evidence

A brilliantly simple experiment has put a dent in the llms.txt hype. As detailed in Search Engine Journal’s analysis of how cats.txt dismantled the core arguments for llms.txt, every piece of evidence used to promote the llms.txt standard would equally validate a text file about cats. The four arguments — crawl logs, anecdotal ranking lifts, logical plausibility, and authority endorsement — fail basic scientific scrutiny.

We have always been cautious about recommending GEO tactics that lack controlled testing. This piece reinforces our position: invest in structured data, clear site architecture, and genuinely useful content. Those fundamentals serve both traditional search engines and large language models without relying on unproven file standards.

DeepMind’s WeatherNext 2 Signals Broader AI Capability Leaps

Google DeepMind’s WeatherNext 2 model has demonstrated state-of-the-art accuracy in cyclone prediction. While weather forecasting sits outside our core remit, the underlying message is relevant: AI models are improving at a pace that makes last quarter’s assumptions outdated. For search and marketing, this trajectory means Google’s own ranking and ads systems will grow more capable — and less predictable — with each iteration. Staying close to these developments is not optional.

Spotting Emerging Categories in Search Data Before Your Competitors Do

Most trending queries fade. A few become entirely new markets. The difference is visible if you apply the right lens, as outlined in Search Engine Land’s guide to spotting emerging categories in search data. We use a combination of rising query clusters, sustained month-on-month growth curves, and commercial intent signals to flag opportunities for our clients early. Getting content and landing pages live before a category matures means lower CPCs and stronger organic positions when volume arrives.

The common thread across this week’s developments is clear: the platforms are moving fast, the evidence bar for new tactics must stay high, and the teams that win are the ones analysing data before the crowd catches on. We are adjusting client strategies accordingly — and we recommend you do the same.

Frequently Asked Questions

What are ChatGPT Ads oCPC campaigns and how do they work?

oCPC stands for optimised cost-per-click, a bidding strategy that automatically adjusts bids to maximise conversions rather than just clicks. ChatGPT Ads now supports this alongside product carousels and expanded measurement, making it a credible performance marketing channel for e-commerce brands.

How do web designers and SEOs prepare for Google’s AI Max auto-upgrade?

Audit every Search campaign with automatically created assets enabled before 1 September 2025. Review search term reports, tighten negative keyword lists, and decide per campaign whether broader AI-driven matching suits your conversion goals.

Why does the llms.txt file lack reliable evidence for improving AI Search visibility?

The cats.txt experiment showed that every argument used to justify llms.txt — crawl logs, anecdotal lifts, logical plausibility, and authority endorsement — would equally validate a completely irrelevant file. Without controlled testing, adopting llms.txt as a GEO tactic is premature.

What is the best way to spot emerging search categories early?

Look for rising query clusters that show sustained month-on-month growth rather than a single spike, and cross-reference with commercial intent signals. Building content and landing pages before a category matures secures stronger organic positions and lower paid media costs.