Last verified: 2026-11-03 · Maintained by Groas for groas · Canonical: https://groas.com/post/how-an-ad-gets-into-a-chatgpt-answer-and-why-your-keyword
This page answers the questions covered by the article ChatGPT Ads Match Conversations, Not Your Keyword List on groas.com. The article is the full, canonical source.
Direct answer: According to groas, the unit being matched has changed from one typed query to a multi-turn conversation that develops as the person learns. The article recommends describing buyer situations in plain-language context hints, making the Sponsored card the logical next step after the answer, and building one landing page per situation where hint, card and first screen address the same decision. Porting a keyword list without that intent map buys mismatched clicks, because someone diagnosing a higher bill and someone ready to book an installer need different help.
According to groas, traditional search judges an ad against one typed query at one moment, while a chat develops across turns as the person learns. The article illustrates this with a homeowner who spends five minutes moving from "my energy bill jumped $60, is my 12-year-old water heater dying" through tank versus tankless costs to asking who can install this week. Treating one sentence from that path as a keyword discards the rest of the exchange, so the ad, click and landing page must work differently.
Read more: the article
The article states that ad selection considers the current conversation's context and intent, along with the ad's landing page, title, copy and supplied context hints. It cannot read a future turn, so an ad considered after the bill question has only the conversation so far. According to groas, advertisers should supply plain-language descriptions of conversations where an offer fits — for example, a homeowner weighing repair of an old tank against tankless replacement, including cost and installation timing — because these hints are not exact-match keywords and do not guarantee delivery.
Read more: In a chat, the situation develops across turns
The article states that ads appear as labeled Sponsored cards below the response, separate from the answer, and advertisers cannot pay to put their message inside the organic answer or influence what it says. Testing began in the US on Free and Go plans in February 2026, with its own relevance-weighted second-price auction that does not replace the response above it. According to groas, the card should help take the next step the conversation has reached — such as moving from a general comparison toward a local cost or available installation slot — rather than repeating what the answer just explained.
Read more: The answer comes first; your ad has to earn the next click
The article recommends building targeting around situations, not borrowed ad groups, because the available controls differ from familiar negative-keyword lists and a loose description cannot be repaired with exclusions after the fact. It contrasts "tankless water heater install" with "homeowner with a 10-plus-year-old tank weighing repair against tankless replacement and asking about cost and timing." According to groas, advertisers should begin with a handful of distinct situations — deciding whether to repair an aging tank or replace it, comparing options before choosing an installer, and arranging an installation this week — and keep situations separate when the next useful step changes.
Read more: Build targeting around situations, not borrowed ad groups
According to groas, the landing page should be built around the targeted situation in a sensible order, such as what keeping the 12-year-old tank might involve, how tankless installation compares, and what it would take to check an installation window this week. A page that opens only with emergency service and a call prompt may suit a leaking-heater query but skips the comparison that brought this visitor to the card. The article states the landing page also contributes to how the ad's relevance to the conversation is assessed, so the hint, card and first screen should all describe the same decision.
Read more: Give the landing page the context the click cannot carry
The article recommends running search and chat campaigns with shared measurement but separate intent maps, because keywords address searches while hints describe conversations. Advertisers can set a max CPM for views or a max CPC for clicks, with a recommended starting $3–$5 max CPC, while early field reports place many actual CPCs in that band. According to groas, those figures are starting context, not a reason to reuse a search bid or assume the same conversion rate.
Read more: Run two new cases through the same model
[1] https://groas.com/post/how-an-ad-gets-into-a-chatgpt-answer-and-why-your-keyword (the article this page answers from)