Prompt Set Design
Fifty or more priority prompts built from the questions your buyers actually ask before purchase.
AI Search Monitoring
Know Exactly What AI Says About Your Brand
A fixed set of priority prompts run across every major assistant on a schedule, recording whether you appear, how you are described, and who gets cited instead.
- 4 assistants tracked
- 50+ prompts per brand
- Monthly reporting
- Competitor share of answer
The Problem
You Cannot Improve What You Have Not Measured
Most brands have no idea how they appear in AI answers. They have anecdotes: someone asked ChatGPT once and it said something odd. That is not a baseline and it cannot support a decision.
A fixed prompt set, run the same way on a schedule, turns AI visibility from a talking point into a workstream with a number attached.
- Prompts built from questions your buyers genuinely ask
- Same set run identically each cycle
- Appearance, position and description all recorded
- Competitor citations tracked alongside yours
What we monitor
- ChatGPT and GPT-powered tools
- Google Gemini
- Perplexity
- Google AI Overviews
- Description accuracy
- Competitor citations
What You Get
Reporting You Can Act On
Not a dashboard of numbers with no next step attached.
Scheduled Runs
The same set executed identically each cycle, so changes reflect reality rather than prompt variation.
Appearance Tracking
Whether you are named, where in the answer, and how that has moved since the last run.
Description Accuracy
What the assistant says about you, flagged when it is wrong or out of date.
Competitor Share
Who is being cited instead of you, on which prompts, and how consistently.
Action Reporting
Every report ends with what to fix next, not just what the numbers were.
How It Runs
From brief to measurable result
Four stages, run in order, each with something you can review before the next one starts.
-
01
Design
Build the Prompt Set
- Buyer questions gathered
- Commercial and comparison prompts included
- Competitor-trigger prompts added
- Set agreed and frozen
-
02
Baseline
Run It Cold
- Every assistant queried
- Responses captured verbatim
- Appearance and position logged
- Competitor citations recorded
-
03
Analyse
Find the Cause
- Gaps traced to sources or structure
- Inaccuracies traced to their origin
- Competitor advantages identified
- Fix list prioritised
-
04
Report
Show the Movement
- Set re-run on schedule
- Change reported per prompt
- Wins and regressions both shown
- Next actions recommended
How We Report
Honest Numbers, Including the Bad Ones
Regressions get reported as clearly as gains. A monitoring service that only shows improvement is a marketing document.
- Fixed, frozen prompt sets
- Identical scheduled runs
- Four assistants covered
- Verbatim response capture
- Competitor citation tracking
- Action-led reporting
- Regressions shown too
- Inaccuracy root-cause tracing
- Next steps every cycle
Related
Pairs well with
Most clients run these alongside AI Search Monitoring & Reporting.
FAQ
AI Search Monitoring & Reporting questions
Something not covered here? Ask us directly and a strategist will reply within one working day.
Fifty or more for a useful picture. Fewer than that and normal model variation drowns out any real signal.
Monthly for most brands. Weekly is worth it during an active campaign or after a significant content or PR push.
Yes, models vary. That is precisely why the set is frozen and run identically each cycle, so variation averages out and real movement becomes visible.
We trace the inaccuracy to its source. It is almost always something on the web saying it, and correcting that source is the fix.
Yes, and it is one of the more useful parts. Knowing who gets cited instead of you, and on which prompts, tells you exactly where to focus.
As a diagnostic, yes. But monitoring only tells you where you stand; changing it needs the structural and authority work alongside it.
Let's Figure Out What Growth Actually Looks Like
We look closely at your website, your industry, and where you currently stand. Then we walk you through exactly what we do, and how much it would cost.