What a sweep means
In AI visibility tracking, a sweep is one complete measurement run: every prompt in your tracked panel is submitted to every monitored engine, every answer is collected and scored, and the results are stored as a single dated snapshot. If you track 40 prompts across ChatGPT, Gemini, Perplexity, and Claude, one sweep produces 160 fresh answers — a full census of how the AI landscape presents your brand at that moment.
The term borrows from monitoring disciplines (a radar sweep covers the whole field, uniformly, on a schedule), and the analogy is apt: the value isn't any single reading but the uniform, repeated pass.
Anatomy of a sweep
A sweep has three phases:
- Collection. Each tracked prompt is run against each engine, the way a real user would ask it. Answers are captured verbatim.
- Scoring. Every answer is read by an LLM judge, which detects brand mentions for you and your tracked competitors, rates prominence and sentiment, and checks claims against verified brand facts for accuracy flags.
- Aggregation. Scores roll up into the sweep's snapshot: mention rate, share of voice, sentiment, accuracy issues, and an overall AI Visibility Score — all comparable to previous sweeps because the panel and method held still.
Why scheduled sweeps beat spot checks
Asking ChatGPT about your brand once tells you what one engine said one time. AI answers are generated probabilistically — the same question yields different answers across runs, engines, and days. Spot checks therefore produce anecdotes: alarming or flattering, but unrepresentative either way. Sweeps fix all three weaknesses of casual checking:
- Completeness. Every prompt, every engine, every time — no cherry-picking the questions you happen to remember.
- Consistency. The same panel measured the same way makes numbers comparable across time. A visibility change between sweeps reflects the world changing, not the method.
- Cadence. Regular snapshots turn measurements into trends — the difference between "an answer mentioned us" and "our mention rate rose for the fourth consecutive week, driven by Perplexity."
Sweeps are what make GEO an accountable discipline: launch a content push or a digital PR campaign, then watch whether subsequent sweeps move — cause, effect, and timing all on the record.
Reading sweep results
Between any two sweeps, the useful questions are directional: Which prompts did we newly win or lose? Did any engine's framing shift? Did a competitor's share of voice jump — and on which prompts? Did new accuracy flags appear? Because each sweep stores complete per-answer detail, movement in a headline number can always be traced down to the specific answers that caused it.
Sweeps in Brandflare
Sweeps are the heartbeat of Brandflare: they run automatically on a schedule set by plan tier, with every answer judged for mentions, sentiment, competitors, and accuracy — the full mechanics are documented in how sweeps work. If you've never measured at all, a one-off free audit is effectively your first mini-sweep: a baseline snapshot of how the four major engines answer your category's questions today.