Skip to main content

Understanding and using Automatic Themes

Set up and manage automatic themes in your Kapiche project.

C
Written by Cameron Parry
Updated over a week ago

Kapiche’s Automatic Themes feature helps you uncover what customers are saying — fast. By leveraging AI, we take the heavy lifting out of manually coding themes, letting you go from raw feedback to structured insight in minutes.


What Are Automatic Themes?

Automatic Themes are Kapiche’s way of turning raw customer feedback into a structured framework — without the need for manual tagging or coding.

We use a combination of bottom-up discovery and top-down organization to help you go from tens of thousands of comments to a clean, usable insight structure.

🔽 Bottom-Up: Discovering Themes from the Data

Kapiche starts by reading each piece of feedback using generative AI to detect key ideas — we call these topics.

  • Topics are then grouped into themes using unsupervised machine learning.

  • These themes represent the most granular, data-driven units of meaning in your analysis — shaped directly by what customers are saying.

🔼 Top-Down: Organizing Themes for Reporting

Once themes are created, Kapiche helps you make sense of them at a higher level:

  • Themes are automatically arranged into theme groups using AI — this gives you a clear reporting structure for tracking issues, opportunities, or business areas.

  • If you have your own multi-level reporting hierarchy (e.g. Business Area → Category), Kapiche can map themes into that automatically too.

This dual approach ensures your insights are both faithful to the data and ready for reporting and decision-making.


🛠 How to Use Automatic Themes

To get started:

  1. Enable AI Themes
    If your site doesn't yet have AI Themes enabled, get in touch with us and we’ll help you activate it.

  2. Fill in Dataset Context
    When creating a new project, provide the dataset context — this helps us tailor the results to your business or use case.

    Dataset Type

    Dataset Description

    (only requried for some dataset types)

    Text Field Descriptions

  3. Generate Themes on Analysis Creation
    When creating a new analysis, the option to “Generate AI Themes” should be automatically enabled.

  4. Explore in Theme Builder
    After the analysis runs:

    • Head to the Theme Builder to explore the results.

    • You’ll see key statistics (1) and a summary of each top-level Theme Group (2)

    Just like with manual themes, you can:

    • Rename themes

    • Move them between groups

    • Merge or delete them as needed

  5. Explore in the Workbench

    Once you've evaluated your theme framework, go to the Workbench to start finding insights.


🔗 Merging Similar Themes

Kapiche aims to give you themes that are both accurate and readable, though we lean slightly toward precision. That means some similar ideas may show up as separate themes.

You can consolidate and streamline your framework by:

  • Reviewing the themes in Theme Builder

  • Merging any that you feel are too close or redundant
    (See the Theme Builder Guide for step-by-step instructions.)


🔄 What Happens When New Data Is Added?

When your dataset grows, Kapiche automatically processes the new feedback:

  1. New topics are detected in the fresh feedback.

  2. If a new topic matches an existing theme, it gets added there.

  3. If not, we evaluate whether it deserves its own theme. To keep things focused, only new themes that account for at least 1% of verbatims are created.

  4. New themes appear in the Theme Builder in a New Themes folder for you to choose how and when to update your existing theme framework

This keeps your theme structure current without becoming noisy.


Questions? 🤔

That's it for this article. If you have any questions about automatic themes in Kapiche (or anything else!) you can get in touch with us any time by hitting the blue chat button to your right 👉

Did this answer your question?