Prompt Suggestions — Prominara Documentation

Prominara reads your site and drafts the questions your buyers ask AI, then assembles them into a balanced set sized to your plan — instead of handing you a long list to grind through.

Where suggestions come from

Site signals — topics, entities, and competitors detected during scans.
GSC queries — when Google Search Console is connected, real searches your site already ranks for are seeded as candidates.
AI generation — GPT-5-mini drafts questions that make an AI list companies (“best CRM for startups”, not “how does CRM work”).
Templates — zero-cost industry templates as an English-language fallback when AI generation is unavailable.

Template fallback is English only, by design. The old translated templates were software-shaped whatever the industry — a Spanish restaurant was offered “mejor software de restaurantes”. Questions that are wrong about your category are worse than none, so for other languages the wizard asks you to retry generation or write your own instead of filling the screen with nonsense. AI generation itself works in every supported language.

How the set is balanced

A prompt library is a sample of how buyers search, not a keyword list — so allocation happens on two axes at once.

Question type — Discovery gets the largest share, with the rest split between questions about your brand and comparisons with rivals. Each is a separate reading; only Discovery counts toward your headline visibility rate.
Buying moment — inside Discovery, slots are spread across the journey: exploring a problem, looking for options, comparing choices, ready to buy. Weights lean toward the earlier moments, where most searching actually happens.

Small plans still lead with Discovery: when there aren't enough slots to fill every panel, brand and comparison slots are given up before Discovery ones. And with no known competitors, the comparison slots fold back into Discovery rather than inventing a rival name.

How opportunity score is computed

Within each type, candidates are ordered by an opportunity score so the strongest question fills each slot.

Each suggestion gets an opportunity score from three ingredients:

Query type baseline — organic prompts start higher than competitive ones, which start higher than branded ones (branded queries echo your name back and aren't a real visibility test).
GSC overlap — when a suggestion overlaps a real Search Console query your site already gets impressions for, it scores higher. No GSC connection? The score still works; GSC just adds confidence when it's there.
Signal fit — how well the prompt matches the topics, product categories, use cases, and audiences detected on your site.

The same opportunity score orders results across the validation wizard, onboarding, and the API — so the “best next question” looks the same everywhere.

Picking from the list

Start from the recommended set — it's already selected when the workbench opens. “Use recommended set” restores it at any point.
Read each row for whether it sounds like your customers, and edit the wording in place when it doesn't. Anything you edit is kept exactly as you wrote it.
Watch the sidebar rather than the list: it shows slots used and which buying moments the set still misses.

Why Discovery questions matter most

A prompt like “Is Acme good?” will always echo your brand back — that's not visibility. Questions that elicit a list of companies (e.g. “top project management software”) are the ones where Prominara can tell whether AI assistants cite you or your competitors. That is why Discovery takes the largest share of the set and is the only type behind the headline rate — brand and comparison questions are still worth tracking, they just answer a different question.