Digital advertising is usually designed to take a person to a landing page, a cart, or a form. But as people increasingly search, compare, and ask questions in AI environments, the transition may no longer be a click to a website. It may be a conversation in which a customer asks about a product, terms, fit, or use case. That does not make established marketing principles disappear; it makes the quality of information and the coordination among marketing, sales, and operations more visible.
What is new?
On 16 September 2026, OpenAI announced AI-powered advertising experiences for ChatGPT. According to the announcement, the company is testing Sponsored Agents: after clicking an ad, a user can choose to start a clearly labelled conversation with a business-sponsored agent. OpenAI says that this conversation is separate from ChatGPT’s independent answers and from the user’s original ChatGPT conversation; the test is currently available to select advertisers in the United States.
In the same announcement, OpenAI said advertisers can use natural-language prompts in ChatGPT Work with the Ads Manager plugin to create, update, and analyse campaigns. Ads Manager also offers assistance with suggested copy and imagery based on a landing page and campaign objective, while advertisers retain the ability to review and decide whether to use suggestions. OpenAI also announced initial integrations with HubSpot and Shopify; the Shopify app is described as available in the United States, with international availability in markets where ChatGPT Ads are available planned from 23 September. These are OpenAI product statements, not evidence that every business will immediately achieve better results.
Source: OpenAI, “Reimagining advertising with AI”, 16 September 2026.
Management implication: advertising must answer the next question
The management interpretation is this: if an advertisement leads into a conversation, its content cannot optimise only for clicks. It needs to open a topic the business can answer consistently afterwards. A message that is too broad may attract many questions, but if information on pricing, service scope, terms, stock, or returns is unclear or inconsistent, the conversation will expose the gap between the promise and operating capability.
Before testing any conversational advertising format, a business should therefore prepare a “foundation answer set” for the questions likely to follow an ad. This is not a rigid script. It is a verified list covering: which customers fit; what problem the product does and does not solve; price or the quotation approach; evidence that can be shown; limits and conditions; and when to hand the matter to a person. Marketing owns the clarity of the promise; sales and customer service help validate real situations; legal or compliance teams review sensitive claims.
Three practical changes to prepare
| Change | What to do | Risk if ignored |
|---|---|---|
| From keywords to intent | Group likely customer questions by need, context, and uncertainty. | An ad attracts relevant views but leads to generic answers. |
| From landing pages to structured information | Standardise product information, pricing, terms, FAQs, and evidence sources. | Different channels give different answers or retain outdated information. |
| From media metrics to transition quality | Track follow-up questions, hand-off rate, response time, and post-conversation outcomes. | The team optimises CTR without knowing whether conversations create real demand. |
For example, a retail-management software business may run an ad about “reducing inventory loss”. An interested person may immediately ask how many locations the software suits, whether it connects to their existing scanner, how long implementation takes, who helps when legacy data is wrong, and how cost changes with user numbers. A good answer does not have to promise everything; it must answer the verified part, state conditions, and hand off at the right time to an adviser who has context.
Do not confuse a new tool with a new strategy
OpenAI’s announcement shows one development worth watching: advertising channels may become more closely connected to the place where customers consider and ask questions. But availability still depends on market, account, and testing stage. A business does not need to rebuild its whole advertising budget simply because a new feature appears. A more sensible step is to examine its existing foundations: is product information clear enough; does the team agree on what it may promise; and can it measure what happens after customers ask a follow-up question?
If these three foundations are weak, an AI-opened conversation will only reveal existing weaknesses more quickly. Conversely, if a business has standardised information and defined when automation answers and when a person is needed, new formats can become a valuable learning channel: not merely showing which ad was viewed, but showing what information customers still need before moving toward a decision.
