Short answer
Buyers don't type keywords into AI assistants; they describe their situation. Turn each keyword into questions by adding five context slots: company size, industry, role, current tools and a constraint or deadline. Tag each question by stage (discovery, shortlist, decision), validate against real search terms and sales calls, then use the decision questions for ChatGPT ads and Google AI campaigns and the full set to track whether AI names you.
Why keywords stop working
A search keyword is a compressed guess at intent. In an AI assistant the buyer writes the intent out: "We're a 200-person SaaS company on AWS with a SOC 2 audit in Q2. Which compliance tool can we roll out in six weeks?" That sentence carries size, industry, stack and deadline, the exact qualifiers ABM teams pay data vendors for.
ChatGPT ads match on that conversation. Google's ads in AI Overviews come from existing Search, Shopping and Performance Max campaigns using broad match or keywordless targeting, which read the longer query too; AI Mode ads have been in testing in the US. Planning in questions lets you meet both.
The five context slots
| Slot | Example | Why it matters |
|---|---|---|
| Company size | "a 200-person company" | Separates enterprise from mid-market fit |
| Industry | "SaaS", "a 3PL", "a mid-size insurer" | Only your segment asks it |
| Role | "as a security lead" | Tells you which committee member is asking |
| Current tools | "on NetSuite", "integrates with Jira" | Integration questions are high intent |
| Constraint | "audit in Q2", "migrating this year" | Trigger events mark in-market accounts |
Tag every question by stage
- Discovery: "What should I look for in…?" Good for content and AI visibility; weak for ads.
- Shortlist: "Best … for a company like ours", "Which … integrates with…?" The core of ChatGPT ad groups.
- Decision: cost, rollout time, "worth it if…". Highest intent, smallest volume. Fund these first.
Use one list three ways
| Where | How the questions are used |
|---|---|
| ChatGPT ads | One ad group per question theme; context hints written as buyer questions. How context hints work |
| Google Ads | Themes for AI Max and broad match, and landing page copy that answers the long query |
| AI visibility | A fixed prompt set to check whether ChatGPT, Gemini and AI Mode name you. Which prompts to track |
Validate before you spend
- Keep questions that match wording in sales call notes or support tickets.
- Check search term reports for the long variants that already convert.
- Drop questions your best customers would never ask, however much volume they promise.
- Write negatives for the conversations you don't want with the negative phrases generator.
Research on generative engines also shows that answers reward pages that address the full question with specific facts, so the same list should shape your landing pages and FAQs.
Questions
Do conversational queries replace keywords?
Not in Google Ads, where keywords and broad match still run the auction. They replace keywords in ChatGPT ads, which match the conversation, and in AI visibility tracking.
How many questions per keyword?
Eight to twelve is enough. More variants of the same intent add little; new context slots add a lot.
Can I buy ads inside Gemini?
Google sells ads in AI Overviews through existing Google Ads campaigns, and AI Mode ads have been in testing. Check Google's current documentation for Gemini app placements before planning on them.
Where do the best questions come from?
Sales call notes, support tickets and search term reports, in that order. They use the buyer's own words.
Facts checked against the sources below on 19 October 2026. Ad platforms change quickly: if something here no longer matches what you see, tell us and we will update it.
