In partnership with
Happy Monday folks!
Everything you know how to measure happens after the click. Sessions, scroll depth, time on page, the form. Twenty years of craft, all of it downstream of a link.
Now imagine the link goes nowhere.
On Wednesday OpenAI introduced Sponsored Agents: click an ad inside ChatGPT and you can start "a clearly labeled conversation with a business-sponsored agent in ChatGPT", held apart from its own answers. It is in test with selected US advertisers, alongside an ads manager that builds campaigns from plain language and drafts copy and imagery off your landing page. HubSpot is the first CRM partner. Shopify is the first ecommerce partner, with US merchants on it now and other markets from September 23.
Wayfair is in the pilot and said so. "We're testing at limited scale with guardrails around product accuracy, transparency and service handoffs," Morgan Brown, its director of paid media, told Digiday. Guardrails is the tell. He is describing a salesperson he cannot hear.
The buy side is less polite. "While they've been a hot new platform for advertisers, the ROI just hasn't been there for the most part," said Shamsul Chowdhury, svp paid media at Zeno Group. The Register was blunter: slop ads served in slop responses.
Somebody owns that conversation log and it is not you. Ask who reads it, and what happens when the agent promises a delivery date.
Today’s sponsor:
Blu Dot surpasses 2,000% ROAS with self-serve CTV ads
Home furniture brand Blu Dot blew up on CTV with help from Roku Ads Manager. Here’s how:
After a test campaign reached 211,000 households and achieved 1,010% ROAS, the brand went all in to promote its annual sales event. It removed age and income constraints to expand reach and shifted budget to custom audiences and retargeting, where intent was strongest.
The results speak for themselves. As Blu Dot increased their investment by 10x, ROAS jumped to 2,308% and more page-view conversions surpassed 50,000.
“For CTV campaigns, Roku has been a top performer,” said Claire Folkestad, Paid Media Strategist, Blu Dot. “Comping to our other platforms, we have seen really strong ROAS… and highly efficient CPMs, lower than any other CTV partner we've worked with.”
Using Roku Ads Manager, the campaign moved from a pilot to a permanent performance engine for the brand.
TALK ON THE MARKET
Alyson Griffin, head of marketing at State Farm, on what winning inside an AI answer actually looks like, to Digiday:
If a choice is given, or if State Farm is offered up, we want the person to at least pause for a moment. That's a win.
She is not buying her way into the model. State Farm has just rolled out another Patrick Mahomes campaign, and Griffin says AI search made that work more important, not less. Brand as insurance against a ranked list.
Want to see case studies of AI-powered campaigns?
NUMBERS MATTER
26% of organizations already running agentic AI, a quarter cannot detect unauthorized AI agents operating inside their own walls, per EY's AI Risk and Governance Survey of 202 senior AI decision-makers at US companies above $1 billion in revenue. Ninety-eight percent have a governance policy. Forty-seven percent have skipped it for an urgent deployment.
Marketing usually owns the most agents and the shortest inventory of them.
DAILY PROMPT
The full Competitive Ad Analysis Framework from the hub. Point it at whoever is about to be sitting next to you inside a sponsored conversation:
Run a competitive intelligence analysis of the advertising below and return actionable insights for our marketing team. Attach ad screenshots, ad library exports, landing page captures, or spend estimates, and use web research to pull current live creative from public ad libraries where it is available.
## Competitor Information
- Competitor Name(s): [LIST COMPETITORS]
- Industry/Category: [YOUR INDUSTRY]
- Our Target Audience: [DESCRIBE AUDIENCE]
- Campaign Duration (if known): [TIME PERIOD]
## Ads to Analyze
For each ad, work from the attached files or the details below:
- Platform: [FACEBOOK/GOOGLE/LINKEDIN/TIKTOK/OTHER]
- Ad Copy: [PASTE EXACT AD TEXT]
- Visual Description: [DESCRIBE IMAGE/VIDEO]
- Estimated Spend Level: [HIGH/MEDIUM/LOW]
- Call-to-Action: [DESCRIBE CTA]
If an ad set is incomplete, retrieve what you can from public ad transparency sources and flag anything you inferred rather than observed.
## Analysis Framework
For each ad, provide:
1. **Core Message Strategy**: What is the primary value proposition and emotional hook?
2. **Audience Targeting Signals**: What audience segments appear to be targeted based on messaging, visuals, and placement?
3. **Creative Elements**: What makes this ad stand out? (Color, copy style, format, social proof, urgency)
4. **Conversion Mechanics**: What is the likely conversion path? (Landing page type, offer structure, friction points)
5. **Competitive Positioning**: How do they position against alternatives or status quo?
6. **Estimated Performance Indicators**: Based on creative quality, placement, and relevance, what metrics might this achieve, and on what basis?
## Synthesis & Recommendations
After analyzing all ads:
- Identify 3-5 common patterns across their campaigns, including creative variants that suggest active testing
- Highlight 2-3 gaps in their strategy we could exploit
- Recommend 3 specific creative angles we should test, with hooks and CTAs written out
- Suggest messaging frameworks that differentiate us
- Propose 2-3 audience segments they are underserving
Close with a prioritized test plan for our next campaign sprint: what to build, what to measure, and what evidence would prove or disprove each hypothesis.Run it the week a platform pitches you conversational ads, and add one question of your own to the output: after the click, who answers, what may they promise, and where is that conversation logged. The answer you cannot give is the one to fix first.
CMO CORNER
Influence doesn't mean interference.
You cannot edit what a model says about you. You can change what it reads. Same discipline you use on a team that has outgrown your desk: stop rewriting the work, start improving the inputs, then let it run.
The brands that will do well inside a sponsored conversation are the ones whose product facts are already clean enough to survive being paraphrased by a stranger.
We provide consulting services for CMOs on AI strategy for marketing teams through our Teamless offering.
TOOL UPGRADE
If the ad ends in a conversation, three places to put your money:
Chipp — best for standing up a branded AI agent your customers can actually talk to, before a platform builds one for you. Try Chipp here →
Replo — best for ecommerce teams still building the page a click lands on, and testing it properly. Try Replo here →
Unbounce — best for finding out whether a page or a conversation converts better on the same spend. Try Unbounce here →
→ We have 300+ curated, filtered, and categorized AI tools in our directory. Browse to find the best ones that help your marketing team.
LEARN THIS
Every vendor in this space will be in your inbox by Friday: AI Vendor Selection Framework for Marketing Leaders.
It is built for the meeting where the pilot sounds free: what data leaves the building, who owns the output, what happens when you stop paying. Read it before the demo, not after the contract.
→ We publish workshops, courses, guides, manuals, and more on AI marketing for free. Browse our learning resources for free.
BRAND SLOP
Molly Tranchin shot three videos for the underwear brand Eby, under contract, posing with her arms across her chest. According to her lawsuit, what went up instead was a deepfaked version of her, as the Guardian reported, altered to expose her through a sheer bra.
Lifestyle blogger Emily Schuman found herself in a sponsored post for a telehealth company she had never heard of, holding weight-loss drugs, built from a selfie she posted years ago.
Somewhere in your contract with a creator there is a clause about approvals and a clause about edits. Go read the second one. It was written when editing meant a trim.