How often does the free trial scan?
The trial (2 days of full AI visibility) runs 2 scans. The first starts the moment you click Start first scan. Each later scan runs about 24 hours after the one before, and never sooner than 20 hours, so a signup late in the evening still gets a real second day rather than a second scan minutes later.
Each scan asks the prompts you chose, up to 25, on ChatGPT, Perplexity, Claude and Google AI Overviews. When the last scan has finished, scanning pauses. Every answer stays in your workspace. Until then, the app shows when the next scan runs.
On the trial the questions are fixed once the first scan starts, so every trial scan asks the same set and the two days can be compared. Changing questions is part of a plan.
How often does a paid plan scan?
As often as the cadence you chose on the price card: daily or weekly. Daily, the default, is one answer per tracked question, per engine, per UTC calendar day. Weekly is one a week, with a lower brand fee and a lower price per question.
- At each scheduled run, at a fixed time, the run picks up every question your plan tracks for each brand.
- Each question goes to every engine on your plan, the ones you picked from ChatGPT, Perplexity, Google AI Overview and Claude.
- Each question and engine pair gets exactly one job per run, so a retry or a second trigger can never ask it twice or bill it twice.
- The answers, mentions and citations land in your AI Visibility report, and the app shows the last run and the next one.
Our own measurements show why daily is the default: Perplexity changes about 27% of its answer set from one day to the next, and Claude about 9%. The figures and the reasoning are on Pricing and the engine trackers.
Can I run a scan now instead of waiting?
No. Scans run on your plan's schedule; there is no manual run. Every tracked question is asked on every engine at each scheduled run, for each brand in the workspace.
For answers more often, pick a faster cadence on the price card. On the trial, the scans run on the schedule above.
How often can Proofsource detect day-to-day changes in AI answers?
As often as your cadence asks: once a day on a daily plan, once a week on weekly. A change only counts when it is larger than normal variation: every rate carries a 95% interval, and an alert fires only when the old and new intervals stop overlapping.
- Answers, mentions, sentiment and citations: every scheduled run.
- Competitor changes (a competitor edits a page that matters) and New sources (a new page names you, a competitor or your category): checked daily.
- Reviews: review-site ratings and mentions are read weekly.
How long until the answers appear?
The first scan is asked straight away on every engine and usually completes within a few minutes; the page fills in engine by engine as answers arrive. On the daily run, Claude's answers are collected in a batch and can land later in the day than the other engines. A failed read is recorded as a failure and retried, never counted as an answer that left you out.
