A query universe is the curated set of 50 to 150 conversational buyer questions that an AEO program measures itself against. It is the single most important asset of an AEO program — without a stable query universe, week-over-week comparisons are meaningless and metrics drift with sampling whim. Building the universe is also the work that forces the AEO program to declare what it is actually trying to win.
What a buyer question looks like
Buyer questions are full conversational prompts that mirror how a real buyer would type into ChatGPT or Perplexity, not search-engine-style keywords. 'Best CRM for a 20-person SaaS team' is closer than 'best CRM.' 'How does Notion compare to Confluence for an engineering team that values markdown' is a buyer question. 'Notion vs Confluence' is a search query. The difference matters because conversational prompts surface a different brand list than keyword prompts.
Where to source candidates
Five sources combine to produce a strong universe. Sales call transcripts, where prospects literally state the question they researched. Customer success notes about evaluation criteria. Competitor comparison articles, scraped for the implicit questions they answer. Reddit threads in your category, especially the question titles. Public Q&A platforms (Quora, AskHubSpot's community). Pull 200 candidates from these, then narrow to the 75 that most directly map to a purchase decision.
How to balance the basket
Three buckets matter. Awareness questions — 'what is the best X' — give the broadest mention-rate signal but are the hardest to win. Comparison questions — 'X vs Y' — are the highest commercial intent. Use-case questions — 'best X for Y' — are the easiest to win because the field is narrower. Aim for roughly 30 percent awareness, 40 percent comparison, 30 percent use-case in the universe. Skewing too far toward awareness leaves you measuring a thing you cannot easily move.
Stability matters more than perfection
Once you set the universe, do not edit it for at least a quarter. Adding a question after the fact (or rephrasing one) breaks the time series and makes share-of-voice movement uninterpretable. The discipline is to lock the universe and accept that some questions will turn out to be wrong picks; you fix them at the next planned revision, with a clear before/after note in the report.
How AskRanker uses the universe
Every AskRanker scan is anchored to your declared universe. We sample each question 25 to 50 times across each model, recompute share of voice and per-question mention rate, and report deltas to the previous scan. The universe also feeds the simulator: when you propose a page edit, we predict the lift across exactly the question basket you picked, not against an average we made up. The universe is the contract between the measurement program and the team's strategic priorities.