AI advertising is expanding faster than the evidence needed to justify its budgets.

OpenAI is adding visual ads and measurement connections as Google urges advertisers to fund demand more flexibly. But delayed outcome signals and weaker click-based attribution make measurement discipline essential: advertisers need to distinguish available demand from proven incremental growth.

Key Takeaways & Market Facts

2024; early 2025Google:Google’s Meridian was introduced in 2024 and made available to all marketers and data scientists in early 2025.[1]
38%Google:The report on Google’s demand-led budgeting presentation states that 38% of retail queries exceed eight words.[1]
>1 billion monthly usersGoogle:The report on Google’s demand-led budgeting presentation states that AI Mode has more than 1 billion monthly users.[1]
0.2%–0.3%PureOne Golf:PureOne Golf’s website conversion rates hovered at 0.2%–0.3% during its struggles with its previous agency.[1]
•OpenAI:OpenAI plans to begin testing visual ads during ChatGPT image generation later in the article’s publication month with a group of U.S. advertisers.[1]
•OpenAI:OpenAI is adding Hightouch, Tealium and LiveRamp integrations to help businesses send conversion data into ChatGPT Ads.[1]

Key trends

Budget flexibility needs evidence-based guardrails
Google’s demand-led budgeting pitch collides with the agentic spending “proof gap”: automated buying can commit money before outcomes are measurable. Meridian and OpenAI’s measurement partnerships point toward broader evaluation, but advertisers still need explicit limits on spending while evidence is pending.
Discovery is moving beyond keywords and visits
Longer AI queries, third-party influence on AI answers and less visible referral paths challenge traffic-led planning. Brands should test visibility around customer decisions and fund learning capabilities rather than assume that visits capture all marketing impact.
Alternative environments compete for performance budgets
ChatGPT’s ad expansion, newsletter sponsorships and the debate over the open web’s prospects suggest a wider set of options for advertisers. Publishers and ad-tech vendors will need credible measurement and differentiated audience access to turn experimentation into recurring spend.

By topic

Measurement & Budget Governance

An analysis of agentic buying argues that campaign budgets and pacing do not resolve the financial exposure created when spending precedes reliable outcome evidence.[1]
PureOne Golf’s attribution case illustrates the disconnect between reassuring agency reports and website conversion rates that hovered at 0.2%–0.3%.[1]
Google’s open-source Meridian framework offers an MMM approach to assessing paid search’s incremental contribution rather than assigning credit solely to the last click.[1]
Google advocates fully funding successful campaigns through demand-led budgeting, while the commentary notes Google’s commercial incentive to increase advertiser spending.[1]

AI Search & Growth Planning

The 2027 planning guidance recommends scenario-based capability funding and bounded experiments rather than relying on last year’s traffic to set budgets.[1]
A B2B SaaS AI visibility audit prioritizes third-party websites and bottom-of-funnel prompts to identify gaps near purchase decisions.[1]
AI search’s shift toward detailed, personal questions gives brands a reason to address specific customer circumstances rather than only short keywords.[1]

Advertising Platforms & Publisher Monetization

OpenAI is expanding ChatGPT Ads with a visual-format test, additional measurement integrations and brand suitability tools.[1]
Substack’s sponsorship expansion and beehiiv’s programmatic ad network highlight advertising’s growing role in competition between newsletter platforms.[1]
A Mobile Dev Memo discussion with Gareth Glaser explores the bull case for the open web in performance marketing rather than documenting a confirmed resurgence.[1]
The YouTube promotion guide recommends recurring, themed video series as one way to build channel views.[1]

Recommendations & Agentic Commerce

An overview argues that longer user histories can improve recommendation relevance, although the supplied examples provide no quantified performance gains.[1]
Mobile Dev Memo frames travel as a potential agentic-commerce opportunity, but the supplied subscription excerpt provides no substantive evidence to assess the thesis.[1]

What to watch

Whether ChatGPT’s U.S. visual-ad test produces evidence of incremental outcomes beyond attributed conversions.
How advertisers cap spending committed before reliable sales evidence arrives.
Whether AI visibility audits and newsletter sponsorships translate into measurable pipeline or sales.