A complete guide to how Geonos builds and tracks your topics and prompts automatically · and how to add your own on top whenever you want to.
Visibility metrics · Opportunity score, Visibility Score, mention rate · are only as good as the prompts behind them. That's why Geonos doesn't hand you a blank page: every account starts with a real set of topics and prompts already built and tracked automatically, grounded in your actual products and real demand data · never a generic template, and never something you have to assemble yourself from scratch. Prompt research then gives you full visibility into exactly what's being tracked, which engines it's asked on, and how often, plus full control to adjust it whenever you want.
Prompts are grouped under topics: a tightly scoped semantic container for related queries.
Core concepts
Topic · A focused semantic category. Think "waterproofing & GTX" (not the vague "waterproofing"), "wide-fit sizing," or "daypacks & thru-hiking gear." Topics should be specific and intent-rich, not a generic product category.
Prompt · A natural question a real buyer would ask, grouped under a topic. Prompts must be specific and high-intent. The more specific the prompt, the cleaner the insight it produces.
How Geonos builds your topics and prompts
Topics come from three sources Geonos merges into one list automatically: product-line decomposition (each core product/service line becomes a topic directly), demand-based sub-topics (product lines cross-referenced against search-plus-community demand to surface the specific attributes people actually ask about), and competitive-gap topics (topics your tracked competitors already own, worth tracking even at low standalone volume). This runs automatically during onboarding's Topics step, and keeps running in the background afterward · you can add, merge, or retire topics anytime from Prompt research, but you never have to build the starting set by hand.
Prompts within a topic are sourced automatically from keyword-to-question reformulation, real "People Also Ask" results, autocomplete suggestions, and community question threads · never freely invented with nothing to check them against, and never something Geonos leaves for you to write from a blank field.
Designing good prompts of your own
Geonos already follows every rule below when it builds your tracked prompts automatically. They matter most for you when you're adding a prompt on top of what's already tracked:
Do
Phrase prompts the way a real buyer would actually type or ask them.
Ground every prompt in a measurable demand signal · search volume, community volume, or both.
Keep one intent per prompt ("wide fit hiking boots that don't rub," not three questions stitched together).
Track a comparison prompt against a named competitor when one exists, not a generic version of it.
Avoid
Single-word or vague prompts ("boots") · they don't match how anyone actually asks an AI engine anything.
Prompts with zero demand signal behind them · Optimization actions can't score what it can't measure.
Duplicate prompts that only differ by phrasing · they split your data instead of strengthening it.
Letting a topic sit empty · an unused topic with no prompts under it won't surface anywhere.
Prompt templates
If you're adding a prompt beyond what Geonos already tracks for you, a few patterns consistently produce trackable, high-intent results: "best [product] for [audience/use case]," "[your brand] vs [competitor]," "[product] vs [alternative material/format]," "how long does [product/component] last," and "is [product] worth the price." Swap in your own topic's terms · these are starting shapes, not fixed text.
Per-prompt settings
Each tracked prompt can be scoped to a subset of the 10 tracked engines (skip an engine if it's structurally irrelevant to a prompt, like a B2B prompt on Alexa for Shopping), tagged for easier filtering later, and marked as suggested vs. manually added so you can tell which prompts Geonos surfaced versus which your team added directly.
Managing your prompt library
Add, edit, or retire a prompt anytime from Prompt research · retiring a prompt keeps its historical data but stops new daily tracking. Use Tags to group prompts across topics for reporting (e.g. everything tied to a specific campaign or competitor).
Using Prompt research
Prompt research is where every tracked prompt lives day to day · each one carries a demand number next to its AI mention rate, so you can tell "nobody's asking this" apart from "people ask this constantly and we're invisible." New prompts are suggested weekly from demand gaps against your current coverage.
Use case examples
Example · Acme Outdoor Co. The topic "Wide-fit sizing" tracks "best trail running shoes for wide feet" and "wide fit hiking boots that don't rub" · both grounded in real demand and both feeding directly into that topic's Opportunity score. When the gap on either prompt widens against a named competitor, it surfaces as a scored Optimization Action, not just a number on a dashboard.
Common questions
How many prompts should I track?
There's no fixed number · track every prompt with real demand behind it. Plan tier controls how many can run Tier 1 auto-deploy actions, not how many prompts you can track.
Can I track a prompt on only some engines?
Yes · per-prompt engine scoping is covered above under Per-prompt settings.
What happens if a prompt has zero search volume?
It still tracks, but it won't rank high in Optimization actions, since Opportunity score weights demand alongside visibility gap. A zero-volume prompt with a real community-conversation signal behind it is treated differently · see How Geonos builds your topics and prompts above.
Where to go next: head back to Prompt research to add what you've learned here, or Optimization actions to see how these prompts already get prioritized.
