Getting started
Collections & scheduling
A collection run is how data enters GEOX. Understanding its two passes explains almost every "why is this number what it is" question.
Inside a run — the two-pass pipeline
Collect
Questions → engines → raw answers
Extract
AI pulls metrics from each answer
Aggregate
Metrics roll up into your charts
- 1
Collect
Every active question is sent to each engine enabled for the brand. The raw answer is saved verbatim — it’s the never-deleted source of truth.
- 2
Extract
A second AI pass re-reads each answer with your brand and competitor context, and pulls structured metrics: mention, rank, sentiment, citations.
- 3
Aggregate
Those metrics roll up into the trends, charts, and tables on every analytics page — sliced by the date range and platform you pick in the top bar.
Why analytics lag a run by a moment
The toast fires when collection finishes; the extraction pass lands shortly after. If KPIs look sparse right after a run, wait a moment and refresh.
Three ways to trigger a run
| Trigger | What it does |
|---|---|
| Run now | Collects all active questions for the current brand. Available on Brand Exposure and on the Brands page. |
| ▶ on a question | Runs just that one question — even if it’s paused. Handy for testing a new question without burning credits on the whole set. |
| Schedule | Fires automatically — daily, weekly, monthly, or a one-time run — at a time in your local time zone. Set from the clock icon on a brand row. |
Every stored answer is tagged Manual or Scheduled so you always know its source; the Questions page shows the tag on each row.
Setting up a schedule
- 1
Open the schedule dialog
Go to Brands and click the clock icon on the brand row you want to automate.
- 2
Pick a cadence and time
Choose daily, weekly, monthly, or a one-time run, and the time of day. Times are interpreted in your local time zone.
- 3
Let the trends build
Each scheduled run adds a new data point. With a daily or weekly cadence, the time-series charts on Brand Exposure and Progress Report become genuinely comparable period over period.
Runs and credits
Runs call live models, so each one consumes credits from your account balance (shown at the bottom of the sidebar and on Settings). Two easy levers keep costs down: pause questions you no longer care about, and disable engines you don’t need on Settings → AI models — fewer questions × fewer engines = a cheaper run.
Start with the three core engines and a daily or weekly schedule. Turn on more engines once you know which questions matter — coverage you don’t read is just spend.