Skip to main content
Skip to main content
All posts

The Ad Clicked. Then It Talked.

OpenAI's Sponsored Agents turn a ChatGPT ad click into a conversation. Here's why that's the direction commerce is going, and why your AEO and product feed have to be ready for it.

Nick Morgan avatarNick MorganFounder & CEOSep 23, 20266 min read
The Ad Clicked. Then It Talked., cover image

On September 16, OpenAI announced Sponsored Agents for ChatGPT Ads. Click an ad in ChatGPT and you can choose to keep going, into a clearly labeled conversation with an agent the business sponsors. Ask questions, get answers, then follow a link to the company's site when you're ready to take the next step.

In the same breath, OpenAI connected ChatGPT Ads to HubSpot, its first CRM partner, and Shopify, its first ecommerce partner. Campaigns can now start where a customer record or a product catalog already lives, not inside a separate ad tool you have to learn.

I've been thinking about what this actually is, because the press mostly called it a new ad format. That undersells it. A new ad format is a new shape of box sitting on a page. What shipped here is something structurally different. The ad clicked, and then it talked back.

What changed, in plain terms#

For most of the internet's history, advertising ended at a landing page. Impression, click, page, form. The buyer did the work of qualifying themselves after they arrived.

Sponsored Agents move the qualification step inside the channel. The user sees an ad, opts into a conversation, and the business's agent answers their questions before any human is involved and before the user ever reaches the site. OpenAI is testing it with select US advertisers now, and the conversation is kept separate from ChatGPT's own answers and from the conversation the user started in 1.

The HubSpot and Shopify pieces matter more than they look. In Shopify, a US merchant can connect an ad account, sync a product catalog, build product-led ads, add a pixel and conversion events, and run the campaign without leaving Shopify 2. In HubSpot, you create ads, track performance, and follow up on leads using the CRM context you already have.

That turns the ad into a qualification and handoff system, not only a placement. The catalog feeds the agent. The CRM catches the lead. The conversation sits in the middle.

Why I think this is where commerce is going#

Here's the bet, stated plainly. Buying is becoming a conversation that starts before the buyer reaches your website, and the companies that can hold up their end of that conversation will take a disproportionate share of the demand flowing through these systems.

Three things point the same direction.

First, the ad surface itself is becoming the store. With Sponsored Agents plus Shopify, the conversation, the catalog, and the path to purchase live inside one surface 3. Search used to send you somewhere. This starts to let you stay and transact.

Second, the qualification is happening earlier. The agent fields the questions a salesperson used to field after the form fill. By the time a person clicks through to your site, they've already asked whether the dining table seats six, how to care for the finish, and whether it fits their space 1. They arrive warmer, or they don't arrive at all, because the agent already told them it wasn't a fit.

Third, the data feeds the experience. The Shopify integration syncs a product catalog and conversion events. The agent is only as good as the catalog behind it. Messy feed, wrong answers, lost conversation. Clean feed, accurate answers, qualified handoff.

Put those together and the direction is clear enough to place a bet on, even this early. Discovery, qualification, and purchase are collapsing toward a single conversation, and the ad is where the conversation starts.

What would prove this wrong? If Sponsored Agents stay a niche test and the conversational layer never broadens beyond a handful of US advertisers, the bet is wrong. I don't think that's where this is going, but it's the honest falsifier. OpenAI hasn't published pricing, targeting, measurement details, or a full rollout date yet 4. The shape is early. The direction isn't.

What this means for you right now#

A lot of companies will read this and think the right move is to start buying ChatGPT Ads. That might be right later. It's not the most important thing right now, and here's why.

The sponsored agent represents your business by drawing on what it can find and understand about you. Your website, your structured data, your product feed, your data feed. That's the material it works from. If the agent can't parse your catalog, or can't find a credible answer to a buyer's question on your site, the conversation goes badly and the handoff never happens.

This is the same substrate that answer engine optimization runs on. The reason a company gets cited organically in ChatGPT, Gemini, Perplexity, and Google AI Overviews is the same reason its sponsored agent will be able to answer well. The structured data, the clear product detail, the entity-level clarity. Get that right and you're stronger on both sides, organic and paid. Get it wrong and you're buying a slot for an agent that can't represent you.

So the practical work doesn't change, even though the channel did.

  • Get cited organically. If you're the dominant answer in your category, the agent has better material to work with, and you're showing up whether someone clicks an ad or not. AEO is the foundation, not the upsell.
  • Clean up your product feed. The Shopify integration makes the feed the creative. Titles, attributes, availability, pricing, variants. The agent answers from this.
  • Fix your structured data. Product, Organization, FAQ, and Offer schema. What the answer engine cites is what the sponsored agent draws from.
  • Make your product detail pages answer questions. The agent fields the questions a buyer asks. If your pages answer them clearly, the agent has something accurate to say.

If you want the one-line version: the answer is the asset. The ad slot is a toll you pay to use it.

The opening most companies will miss#

Here's the thing I keep coming back to. This is enterprise-grade commerce behavior, conversational qualification, catalog-driven personalization, arriving at small and mid-sized business scale. That's the pattern I've watched for twenty years. The capabilities built for the world's largest companies eventually reach the companies that couldn't afford them, and the ones ready for it first take the ground.

The companies that will win here aren't the ones with the biggest ad budget. They're the ones whose site, feed, and data are clean enough that an agent can represent them well, and who are already the answer when a buyer asks the question organically. That work compounds. The ad budget doesn't.

We built Cognos around this thesis before Sponsored Agents existed, because the direction was already visible. The companies that own the answer, in the engines people actually ask, are the ones the agent will represent best when it's their turn to answer back.

The ad clicked. Then it talked. The question is whether your business can hold up its end of the conversation, and that's a question your feed and your AEO answer today, not your media plan.

That's where this is going, and it's moving faster than the last channel did.

About the author

Nick Morgan, Founder & CEO
Nick Morgan

Founder & CEO, Perfectus Labs · United States (remote-first team)

Nick founded Perfectus Labs on a single conviction: the internet became an answer, and the companies named in that answer own the category.

Experience behind this post

  • Founded and exited a full-service agency, then built Perfectus Labs around an AI-native thesis rather than a service retainer.
  • 20+ years scaling companies into eight and nine figures across very different categories and buying cycles.
  • Managed more than $2 billion in marketing investment and helped clients generate more than $20 billion in sales.

Writes reliably on

Where AI-mediated search is taking buying behavior nextWhy a small set of companies will own the answer in every categoryHow leaders should place bets when the discovery channel itself is changingBringing enterprise-grade AI strategy to small and mid-sized businesses

Published under these standards

  • Predictions state plainly what would prove them wrong.
  • Claims are grounded in businesses he has personally built, operated, or advised.
  • Revenue and scale figures reference career totals across companies, not single-year company results.