Segments
A segment groups prompts under a model selection and collection schedule. Use separate segments for research that needs different questions, models, or timing, such as "UK buyer questions" and "Weekly enterprise comparisons".
Create a segment
Create your first segment in the guided project wizard, or open Project Settings -> Segments to add another.
Set:
- Name, so teammates can recognise the research scope.
- Geo and Language, to describe its geographic and language context.
- Models, selecting at least one of the models available on your plan.
- Refresh Frequency, choosing Daily or Weekly.
- Web Search, if your plan allows it and you want supported models to use web grounding.
Save the segment, then add its questions from Answers -> Prompts.
Geography can inform supported site assessments. Language is a context label, not automatic translation. Put location and language requirements into the prompt wording when they matter to the answer.
Choose models and frequency
Start with a small, consistent model set so you can compare results over time. More models produce more responses for each question, and different models can give different answers.
Daily collection is useful for regular checks. Weekly collection suits slower-moving research. Custom arrangements may include hourly collection, but hourly is not a choice in the segment form.
Model availability and web-search access depend on your plan. Use the choices shown in the app and Plan & Usage for your organisation's limits.
Pause and resume
Pause a segment to stop scheduled collection while preserving its prompts and existing evidence. Resume it from Project Settings or the segment detail view when you want collection to continue.
A paused project or prompt can also stop collection. If data is missing, check status at all three levels.
Review a segment
Open its detail view to check scope, models, schedule, prompts, and analytics. Elsewhere in the project, use the segment selector to focus Answers, Performance, or Content Studio on that research scope.
Keep the same segment and model set when comparing periods. Changing them changes the sample behind the metrics.