Skip to content

Methodology Review AI-based analysis with expert judgment.

Check categories against original responses, document your adjustments and assess the results in the context of your study.

A new AI run can produce different results.

Saved results and topic versions capture the state of an analysis. That is why you should check assignments against the original responses before sharing results.

Example

Reviewing and correcting an assignment

This is what an expert review of a single response looks like.

Original response
“The delivery arrived quickly, but I had to call customer service three times to get my refund.”
Open-ended question: What could we improve?
  1. AI assignment

  2. Expert correction

    Created a new subcategory and also assigned the response to it.

Select a category, a sentiment or a step to see the matching passage.

Process

From raw data to reviewed results

  1. Detect topics

    deepsight suggests topics and categories for your texts – based on your own codebook if you wish.

  2. Review assignments

    For each category you see the assigned original responses and check samples.

  3. Adjust categories

    Rename, merge, split or remove topics to fit your research design.

  4. Share results

    Export to Excel and CSV, or as a report and PowerPoint.

Types of evidence

What you can review and document

EvidenceWhat it showsStatus
Original responses per categoryFor every category you see the assigned responses verbatim.Available
Topic versionsChanges from renaming, merging, splitting or removing are saved as a new version.Available
Saved result statesA completed run is kept as a saved state and can be opened later.Available
Versioned analysis profilesStudy context, goals and glossary are saved as a profile, versioned and reused.Available
Export for reviewExport results as Excel or CSV and review them outside the cloud.Available

See the review on your data

We show you assignment, correction and export using an example from your field.