A few weeks ago I wrote about OpenAI opening ChatGPT ads to every U.S. advertiser through self-serve and adding CPC bidding. That was the policy change. Today is the product: OpenAI launched the actual Ads Manager UI inside ChatGPT, with the major holding companies — Dentsu, Omnicom, Publicis, WPP — wired in alongside Adobe, Criteo, and StackAdapt. The revenue target is now $2.5B for the year and $100B annually by 2030.
The May 6 piece argued the brand-side playbook: audit organic GEO before spending a dollar, treat GEO as the prerequisite to paid, build for incrementality. All still true. This post is for the other side of the table — anyone running a sales org whose buyers are increasingly researching inside ChatGPT.
Three things change for sellers when paid placements scale inside the answer layer.
1. The discovery question changes.
A prospect today already shows up partially educated by Google, LinkedIn, peer Slack groups, and what their team has tried. By the end of 2026, add “and whatever ChatGPT showed them” — except now some of that has been paid for. If a buyer arrives saying “I've heard great things about [competitor],” I want to know where. Add one discovery question this quarter: “When you were researching this, what tools did you use?” The answer is now diagnostic.
2. Branded-prompt bidding will become a defensive line item.
The first thing that happened on Google Ads in 2005 was competitors bidding on each other's brand names. Expect the same here. When a prospect types “alternatives to [your product],” somebody will pay to be the answer. Talk to marketing now about who owns the “defensive bidding inside ChatGPT” budget line. The first time you lose a deal to a competitor named in an AI answer your buyer didn't research themselves, that conversation gets a lot more expensive.
3. SDR pre-call research has to include the LLM check.
If your reps prep a call by reading the prospect's LinkedIn and last earnings report, that's no longer sufficient. Five minutes inside ChatGPT — asking the model what it knows about the prospect's company, stated initiatives, and what tools are typical in their stack — surfaces what the prospect themselves was probably told this week. The buyer is taking advice from the same source your reps should be sanity-checking.
The May 6 post worried about the answer layer becoming pay-to-play. Two weeks in, the infrastructure to make that real is shipping. The teams that build a feedback loop between deals lost, what the prospect “knew” coming in, and what the LLMs actually say about their category will spot pattern shifts six months before the marketing team does. Start the loop now.