Prompt tracking

The default prompt mix

Each industry pack ships with a baseline prompt set — usually 30–60 prompts covering the highest-volume buyer queries in the category. You don't need to add anything for the platform to be useful. But adding your own moves you from "good visibility data" to "exact answers to questions you care about."

Adding a custom prompt

Visibility → Prompts → Add custom prompt.

You give the prompt a friendly name, the verbatim text the engine will see, and an optional category tag. Custom prompts run on the same cadence as the rest of your audit.

A good custom prompt:

• Mirrors the way a real buyer would ask the question.
• Avoids your brand name in the prompt itself (defeats the test).
• Is one question, not three.

A bad custom prompt:

• "Tell me about [my brand]" — engines will always cite you on this, so it's noise.
• "What is the best [category] tool, ranked, with pros and cons, and pricing details, and customer reviews?" — too long; engines truncate.

Tracking rank over time

Each tracked prompt has a sparkline showing your rank over the last 30 days, with two horizontal reference lines: the average rank of your top three competitors, and rank 5 (the citation threshold).

Click any prompt for the full timeline view:

• Daily rank per engine (Domination cadence) or weekly (other tiers).
• The actual response text the engine returned.
• The position your brand appears at within the response.
• Whether you're cited (URL link) or just named.

Engine drill-down

The drill-down view stacks engine responses side by side. Useful for spotting patterns: a brand that lives at rank 2 on ChatGPT but is unranked on Perplexity has a citation problem, not a positioning problem.

We store the verbatim response text for 90 days. After that we keep the rank but drop the response body to save storage.

Sharing a prompt with reviewers

Each tracked prompt has a permalink. Share it with a reviewer or stakeholder; they see the historical rank chart, the latest engine responses, and any annotations — without needing a paid seat.

Permalinks honour your role permissions: a Viewer can see, but cannot edit the prompt or trigger a re-run.

Bulk import

CSV upload at Visibility → Prompts → Import. Schema:

name,text,category
"Best CRM small biz","best CRM for small businesses",crm
"Hubspot vs Salesforce","HubSpot vs Salesforce for SMB",comparison

Limits: 200 prompts per workspace on Strategy, 500 on Managed, unlimited on Domination.

Variants and language

A prompt's text is what the engine sees verbatim. To test variants ("best CRM" vs. "top CRM"), add them as separate prompts. We do NOT auto-rephrase, on purpose — small rewording shifts results, and we want you in control of what's being measured.

For non-English audits, the platform supports prompts in any language the connected engines support. The rank logic is language-agnostic.

Auto-suggested prompts

After every audit, the recommender suggests 3–5 new prompts based on gaps and competitor citations. They're optional — accept or dismiss in one click.