For several months OpenAI has been unusually open about its ambition to monetise user prompts with advertising. It spent the last year recruiting Meta's and Google's advertising staff, and its intention to compete directly for consumer intent is obvious. Yesterday's announcement (16 September, Reimagining advertising with AI) opens a new front in the battle for online advertising and protection eyeballs.
The interesting and innovative part is a new, AI-native advertising format. Rather than sponsoring keywords and sending consumers to a website, brands are invited to sponsor and start conversations: a move from cost-per-click to click-to-chat.
The mechanism of OpenAI advertising has clearly been shaped by decades of learning from Google AdWords. A user enters a prompt, and if the context matches an area ChatGPT has an advertiser for, the user sees a labelled ad and is invited to "chat with us". One click and they are in a branded mini-agent conversation inside ChatGPT and, in effect, the consumer starts a direct conversation with the brand.
Why this matters for protection
Protection is bought after, or in preparation for, a life event. A trigger in the consumer's world turns a protection need into a search, and for twenty years the protection industry has paid to intercept that search at Google before passing the consumer to a form.
What changes when the need is typed into ChatGPT instead? Not "cheap life insurance quote" but "we've just had a baby, what do we need to sort out". ChatGPT answers, and somewhere in that answer is a suggestion to talk directly to a life insurer.
It is no exaggeration to say a revolution in monetising consumer intent is coming. Three consequences.
1. Contact rate stops being the metric
Paid search delivered a click from a narrow keyword to a web form, then assumed the consumer knew what was best for them. The sponsored agent receives someone who has already talked through their situation and their objections, and been told by a neutral party that cover is a sensible consideration.
This is not brand bidding on the vague promise of a "free five-minute quote". It is a warm handoff from the most trusted online adviser the consumer has ever had. Every lead in today's market is priced on the assumption that a broker will do a lot of work just to speak to the consumer. ChatGPT's model delivers consumers who open the conversation themselves, full of intent.
2. Who is at risk?
Insurers with no conversational layer. A sponsored agent needs an agentic process that can fetch quotes and complete medical declarations. Today, no UK insurer supports that, end to end.
Aggregators succeeded because the consumer had nowhere else to compare options without the pressure of a telephone sales call (a model latterly hijacked to produce exactly that outcome). When the comparison happens inside the assistant, before the ad, the aggregator's job has already been done.
Lead generators. The cost-per-lead price point that has held for nearly twenty years goes quickly. As consumers migrate to large language models, a form-fill becomes worth a fraction of a completed conversation, and the buyers will know it.
Distributors. Specifically those whose only differentiator is access to pricing get cut out of the conversation. In a short space of time distributors will need direct insurer relationships and products unique to them, or a way of more directly selling their advice.
3. Nobody is talking about compliance… yet
OpenAI's announcement says nothing about regulated advertising categories. That model has not been imagined, let alone built. Wayfair, one of the first brands in the Sponsored Agent test, can hand an agent a product catalogue and let it talk confidently about sofas, but the rules are very different for an insurer.
The moment an agent discusses cover, sum assured, exclusions or price, it is a financial promotion and a regulated conversation is taking place. The record of that conversation is evidence. None of this is OpenAI's problem, and OpenAI's generic agent tooling will not solve it. OpenAI will simply work with the first insurers who can safely take on that risk themselves.
How to win with this format
At a minimum, insurers and distributors should be waking up to the need to build real conversational AI capability. Not a website chatbot that answers questions about opening hours, but an intelligent AI adviser that plugs into quote, underwriting and application systems and can carry a human request through them.
This new advertising format will first launch in the US but, given the speed of AI development, it is safe to assume the UK will not be far behind.
This is the most significant digital advertising moment since the launch of Google AdWords in 2000. Whoever can put a compliant, knowledgeable protection conversation behind that "chat with us" button will access protection needs and intention at a scale we have never seen before.
Alain Desmier is co-founder of Subcontext.