A study of 34,279 ChatGPT queries by Somantra in Australia shows most insurance answers now carry an ad, but few of those ads come from insurers. Here is what insurance marketers can do about it.
When OpenAI switched on advertising in ChatGPT in Australia on 17 April 2026, insurance looked like an obvious fit. People ask AI assistants which car insurance to buy, whether a policy covers flood damage, and how to make a claim. Those are high-intent questions in a category where every new customer is worth a great deal.
Four months later, the ad space beneath those questions is full, but mostly not with insurers. Answer Engine Optimization Research from Sydney-based AEO, GEO plateform Somantra found that 71% of ChatGPT responses to Australian insurance questions carried at least one ad. Across 34,279 queries asked in August 2026, ChatGPT showed 35,290 ads from 371 advertisers. Two thirds of those ads did not sell insurance.
For insurance marketers, the findings point to a channel that is busy, cheap to ignore today, and likely to become harder to enter later. Six practical lessons stand out.
- Measure the answer before buying the ad
The study found that being named in ChatGPT’s answer and being advertised beneath it rarely happen together. In 85% of the responses carrying an ad from Budget Direct, the most active insurance advertiser, the answer itself did not mention Budget Direct. Across 10,832 responses carrying an insurance ad, the advertiser was the first brand named in just 1.5%.
That makes the answer and the ad two separate contests. According to Somantra founder Arun Prasad, insurers need to know how often the answer recommends them before deciding how much to spend beneath it. Somantra tracks this through two measures: a Brand Mindshare Score for how often a brand is present across AI answers, and a Brand Consideration Score for how likely a customer is to choose it based on how the answer frames it against competitors.
“An ad buys presence on the page. It does not buy a place in the answer,” Prasad says. “An insurer that knows where the answer already favours it can spend on the gaps, rather than paying to repeat what the answer is already saying.”
- Defend the answers that already name you
Allianz, NRMA and AAMI were the insurers ChatGPT mentioned most, appearing in 4,710, 4,633 and 3,545 responses respectively. None of them ran ads selling insurance on the day measured. Of the 20 insurers Somantra tracks, only Budget Direct did, holding 61% of all ads that sold insurance.
The result is that the ad slot beneath answers naming a major insurer is open to rivals. Budget Direct’s ad appeared under 21% to 35% of the responses that mentioned each of its eight most-mentioned competitors. The study cannot say whether those placements are deliberately targeted, but for an insurer with strong organic presence, leaving the slot empty means sharing the page with someone else.
- Claim the categories non-insurers have taken
In three insurance categories, the largest advertiser was not an insurer. Form builder Jotform led on pet insurance questions with free pet health record templates. Visa led on travel insurance with travel card messages, and super fund Australian Retirement Trust led on life insurance. Software firms overall held 27.1% of all ads, close to the insurance sector’s 29.7%.
These are buyers asking about cover, served ads that do not offer it. For pet, travel and life insurers, that is unclaimed attention in conversations already about their product.
- Target the moments when people shop
The wording of a question shapes whether an ad appears. Queries using “best”, “top” or “recommended” were about 8 percentage points more likely to carry an ad than other queries, and those naming a state or city about 5 points more likely. Questions beginning with “why” or about claims were less likely to carry one.
The practical lesson is to build campaigns around shopping and comparison moments, and to write ad copy that answers the question being asked. Jotform’s approach is instructive: it ran 258 versions of its ad text, with a headline written for nearly every topic its ads appeared under.
- Send each offer to the page that delivers it
Of the ads from insurers, 55.5% offered a quote. Only 0.1% linked to a web address that looked like a quote page; most went to general product pages. Those pages may still carry a quote button, but every extra step between an offer and its fulfilment is a chance to lose the customer.
Budget Direct offers a model of tighter alignment. Its ads sat in a single motor insurance campaign, with each message theme, such as comparing cover or managing cost, linked to its own dedicated page.
- Get tracking and disclosures right from the start
Advertisers have not yet agreed on how to label ChatGPT traffic. Links were tagged “chatgpt”, “openai”, “chatgptads” and 22 other variations, and 113 of 371 advertisers had no tracking tags detected. Without consistent tagging, marketers cannot compare ChatGPT with search or social, or justify a larger budget.
The study also found that product-document notices, such as “Read PDS & TMD before deciding”, appeared in 39.7% of ads from other financial services advertisers, but none were found in insurer ad text. The check covered ad text only, not the linked pages. Insurers will want their own compliance teams to decide what belongs in ChatGPT ad copy.
An early window
The study has clear limits. Queries were asked on desktop by a user with no history, so a real person’s ads may differ, and the advertiser data reflects a single day. The researchers recorded ad appearances only, with no view of spend, clicks or sales.
Even so, the pattern is clear enough to act on. ChatGPT’s ad space beneath insurance questions is already crowded, and most of the crowd is not in insurance. As tracking, targeting and competition mature, the cost of entering is likely to rise. For insurers, the advantage sits with those who measure their place in the answer now, and decide where an ad would add to it.
