The problem this solves
Traditional share-of-voice tracking covers search rankings and ad presence, but a third surface now shapes deals: what AI assistants say when buyers ask the questions that matter. On that surface most brands have no instrumentation at all. Answers shift between weeks, mentions come and go, and a rival can climb from absent to default recommendation without registering on any dashboard you own. The teams that do check tend to check once, form an impression, and move on.
A useful read has to be repeated and comparative: the same prompts, the same competitors, measured the same way, week after week. That consistency is what turns anecdotes into a trend line - and the trend line is what makes the case for investment, because it shows whether your content and PR work is actually moving the answers. Manual testing cannot hold that discipline; a scheduled mission holds it effortlessly.
How the mission runs
- Fix the panel. The mission locks a measurement panel: your 15 core prompts and the three competitors to score against. Holding the panel constant across runs is what makes week-over-week deltas meaningful, and the brief flags whenever you change it so the trend line stays honest.
- Run the sweep. AI Visibility puts all 15 prompts to the live answer engines and records each response: who is mentioned, who is cited, in what order, and with what framing. Captures are timestamped and stored, building a durable archive of how the answers evolve.
- Score and explain the deltas. Mentions and citations roll up into a share-of-voice scoreboard for you and each rival. The agent explains movement instead of just reporting it - which prompts flipped, what new sources the engines started citing, and the likely cause where the evidence supports one.
- Post the weekly brief. A compact brief posts to your chosen Slack channel: the scoreboard, the week's movers with explanations, and one recommended action when a shift deserves a response. With approvals on, the post waits in your queue; most teams release the first few, then let the schedule run.
The prompt
This is the exact objective the agent receives. Swap the obvious placeholders for your own domain, segment or channel and run it as-is from the console, Slack, or the API.
What comes back
A weekly Slack brief with the AI share-of-voice scoreboard - your brand and three competitors across 15 prompts - plus explained deltas and captured answer excerpts behind any notable move. Over weeks the archive becomes the real asset: a dated, comparable record of who the assistants recommend and how that changed, ready to drop into any quarterly review.
Make it yours
- Split the scoreboard by funnel stage - problem questions, category questions, comparisons - to see exactly where in the journey you lose the assistants' attention.
- Add a fourth tracked competitor temporarily when a newcomer starts appearing in answers, and drop them once the threat reads clearly one way or the other.
- Route the brief to a leadership channel monthly with a rolled-up trend view, while the weekly detail keeps serving the working team.
Frequently asked questions
Why do the numbers move when we changed nothing?
Answer engines update their models and sources continuously, so some wobble is inherent to the surface. The brief separates noise from trend by tracking direction over consecutive weeks, and single-week spikes are labeled as such until they persist. That discipline is the reason the panel stays fixed.
How is share of voice actually computed?
Each prompt's captured answers are scored for brand mentions and citations, weighted by prominence - a lead recommendation counts for more than a trailing aside. Scores aggregate across the 15 prompts into a comparable weekly figure, and the methodology stays constant so weeks can be compared honestly.
Can I audit a week's scoreboard?
Every number traces to stored captures: the prompt, the engine, the full response, the timestamp. The brief links excerpts for the week's notable moves, and the complete archive stays attached to the mission history, so any scoreboard cell can be verified against what the assistant actually said.