Your shopping cart is empty!
Amsive advises treating ChatGPT ads as a pilot and measuring incrementality
Publication date: 25.09.2026
In brief
- Amsive advises running ChatGPT ads as a separate pilot, with no budget shifted over from search.
- OpenAI suggests starting with a maximum CPC of $3 to $5, but no mature third-party benchmarks exist yet.
- The channel should be judged by incremental lift, because last-click conversions undervalue long decisions.
$3–5the maximum CPC OpenAI recommends as a starting range for ads in ChatGPT.
Inna Zeyger, VP of Media Strategy and Growth at Amsive, published a long piece on ChatGPT ads on the agency's site on September 24, 2026. Her main thesis: it is not "search 2.0" but an environment where a person is deciding what to choose.
Ads follow the conversation about a choice
In search you catch a person who has already named a need in the words of a query. In ChatGPT one conversation can travel from trying to understand a problem to comparing vendors, and that used to be split across several separate queries.
So Amsive advises describing the reader's decision instead of the term. Instead of "online MBA, graduate degree" the agency suggests something like flexible online MBA programs for working professionals advancing careers without leaving the workforce. Exact-match logic will not apply here, because context hints work more softly.
On the platform side, Amsive lists Sponsored Agents, meaning direct conversations with an AI representative of the brand, HubSpot and Shopify integrations, product feed support and campaign management with views, clicks and conversions. Users under 18 and Edu-plan subscribers do not see ads, so full reach cannot be counted.
How to measure a channel without benchmarks
Amsive writes that last-click attribution should not be relied on and recommends looking at incrementality: geo holdouts (some regions run without ads), exposed versus unexposed comparisons, and multi-touch conversion paths. The goal is simple: to see whether ChatGPT adds behavior that would not have happened otherwise.
Here is arithmetic on my own example, which is not in Amsive's text. A $4 click and a 2% landing page conversion give $200 per lead, and at 4% it is $100. So the landing page weighs as much as the bid, and Amsive says outright that the destination should continue the conversation instead of sending people to a homepage.
How AI answers shape unpaid brand visibility is covered in the guide to AI Overview and SGE. Paid ChatGPT placements are worth comparing with what you can already measure in Google Ads, where the metrics are long settled.
Who Amsive's piece helps and who it does not
The agency singles out the strongest verticals: B2B and technology, higher education, retail, travel, financial services (with restrictions), healthcare with non-sensitive services, and professional services. For e-commerce it adds that the Shopify integration enables product listings.
Amsive sells media services, so the piece works as a pitch too, and that interest is worth remembering. The article does not specify whether ChatGPT ads are available to Ukrainian advertisers, and I would check that first. In my view the $3 to $5 figure is best read as a ceiling for a first test and nothing more.
What interests me most is when the first independent benchmarks appear: so far all the arithmetic rests on OpenAI's recommendations and the advice of agencies that are also learning on this channel.
