Thematic analysis that reveals the unknown unknowns

We maintain a "golden dataset" that we use as our benchmark. During analysis, if the system isn't confident about its output, it flags that content for our team to review, which helps us track and verify our accuracy claims.

It's pretty straightforward - Level 1 themes are the broadest categories at the top, with Level 2 and 3 providing increasingly detailed subcategories.

You've got flexibility here - you can either upload your own list of themes to track or modify our existing categories to better fit your needs.

We're pretty versatile - we can handle both text and voice data in unstructured formats, plus structured data like numbers and categories. The source doesn't matter as long as it's in a format we support.

For solid thematic analysis, we recommend at least 30 samples per theme. That said, our platform can still identify themes with fewer samples - you'll just want to take those results with a grain of salt.

It depends on how much data you're uploading at once - could be just a few minutes or might take a couple of hours for very large datasets.

Yes, definitely! Check out details here.

Nope! Our platform handles all the preprocessing and cleaning automatically.

Anywhere from a couple hours to about two weeks, depending on your data sources and any security assessments your company needs to do.

Pretty well, though not quite as accurately as English. For English and Latin-based languages, we hit 94.5%+ accuracy for theme analysis and 96.5%+ for sentiment. With other languages, we're around 92% for themes and 95% for sentiment.

Absolutely! You can analyze multiple languages simultaneously, and we'll give you the results all in one language.