Create a Report Using Automatic Classifier
5 mins read - Updated on Nov 15, 2026The Automatic Classifier discovers the themes in your feedback while Kimola creates a Report, so you can analyze a dataset without building a model, labeling examples or training anything first. Kimola groups related feedback into themes, names each theme from what customers wrote and, together with the Sentiment Classifier, shows how customers feel about each one. The Automatic Classifier is available in every package and is selected by default in the Report wizard.
This method fits when you explore a new dataset, have no predefined labels or need quick qualitative findings without preparation. Kimola's automatic theme detection works on unstructured text such as online reviews, open-ended survey responses, social media posts and customer messages, in any industry.
Sign in to Kimola and open Reports, where the report creation area is.
How the Automatic Classifier Works
When the Automatic Classifier is selected, Kimola applies its Dynamic Classification⢠technology to the dataset while the Report is created. It reads the context of each text, finds meaningful patterns and groups related feedback into themes with label names of their own. The themes come from the content itself, not from predefined tags or a fixed taxonomy.
The Automatic Classifier is a content classifier, so it tells what each piece of feedback is about. The Sentiment Classifier, which is selected next to it by default, tells how customers feel. Together they give every theme its own sentiment.
Because the themes are discovered and not chosen, you don't need to know in advance what customers talk about. If you want fixed labels that stay the same from one dataset to the next, use a custom model or a pretrained model instead.
For the survey answers of a mobile network operator, the Automatic Classifier might discover themes such as Coverage in Rural Areas, Billing Errors and Roaming Prices, each with its own sentiment.
Step 1: Add Your Data
Sign in to Kimola, open Reports and add the data you want to analyze in the report creation area. You can upload a file in .xls, .xlsx, .csv or .tsv format by dragging it into the upload area or clicking it. You can also paste a link into the search bar and press Start to create a Report from a link, or choose an integration such as Intercom, Zendesk or Google Sheets to create a Report from an integration. The platforms Kimola can read links from are listed in Supported Platforms for Creating Reports from Links.
[Screenshot: The Reports page with the link search bar, the upload area and the integrations]
Step 2: Choose the Columns or the Report Size
The first step of the wizard depends on your data source. For a file, Sheets & Columns shows a preview of the dataset. Choose the Text Column, which holds the text content to analyze. You can also choose a Url Column, which holds the unique web URL, and a Date Column, which holds the datetime value. Both are optional.

For a link, this step shows the page you added and a slider to choose the Report Size, which sets how many reviews Kimola collects. Confirm that you agree to use the extracted reviews for research purposes only and to follow the terms and copyright rules of the platform.
A new Report is classified with the Automatic Classifier and the Sentiment Classifier by default, so you can click Continue to skip the classifier settings. To check or change them, click Advanced. It adds two steps to the wizard, Classifiers and Structure, between the data step and the Dimensions step.
Step 3: Keep the Automatic Classifier Selected
The Classifiers step asks you to choose one or more classifiers to label your data automatically. Pretrained models are listed under PRE-BUILT MODELS and your own models under CUSTOM MODELS, and a filter box helps you find a model by name. The Automatic Classifier and the Sentiment Classifier are selected when the step opens, and selected models show a check mark. To keep the analysis fully automatic, leave the Automatic Classifier selected and click Continue.
Keep the Sentiment Classifier selected as well. The Overview of a Report needs a sentiment classifier and a content classifier to show the best and worst performing themes.
Keep the Sentiment Classifier selected next to the Automatic Classifier, so the Overview can show the best and worst performing themes.
[Screenshot: The Classifiers step with the Automatic Classifier and the Sentiment Classifier selected]
To use your own label structure instead of discovered themes, choose a custom model or one of Kimola's pretrained models. See Create a Report Using a Custom Model and Create a Report Using Pretrained Models. Custom models are not available in the Starter package.
Step 4: Choose the Structure
The Structure step sets how feedback is labeled in your classifiers. Enable Multi-label Classification is on when the step opens. With it on, Kimola assigns several themes to a piece of feedback that mentions different aspects, and each theme gets its own sentiment, which gives a more complete view of strengths and weaknesses. With it off, each piece of feedback is grouped under one main theme.
A review of a mobile app says that the app is fast and looks clean, but keeps logging the user out. With multi-label classification on, it can receive the themes Speed, Design and Login Problems. With it off, it receives one theme, such as Login Problems.
The setting applies to every classifier of the Report. For the difference in the results, see Multi-label vs. Single-label Classification, and for the Report option, see Create a Report Using Multi-Label Classification.
[Screenshot: The Structure step with the Enable Multi-label Classification switch]
Step 5: Select Dimensions (Optional)
In the Dimensions step, you choose the Dimensions that shape how your feedback turns into insights. A Dimension is a research question Kimola asks of every piece of feedback, such as Pain Points, which asks what goes wrong or creates difficulty. Use All, Drivers, Issues, Actors, Context and Comparisons to browse them by group, and find the ones you selected under My List.

Dimensions don't change the themes the Automatic Classifier discovers. They add a layer of answers on top of the themes and the sentiment. Selecting Dimensions is optional; see Add Dimensions to a Report. Each Dimension you select adds to the queries of the Report, because queries = rows × models × Dimensions.
Queries = rows × models × Dimensions. See How to Calculate Queries for an example.
Step 6: Review and Create the Report
The Review screen summarizes the Report before it is created. Report Title names the Report, and for a link it is filled in from the page. Report Language sets the language of the results, including the names of the themes, and offers English, Spanish, French, German, Portuguese, Dutch and Turkish. Your feedback is analyzed in its original language and is not translated. Source shows where the data comes from, and Required Query shows how many queries the Report will use. See Language Options in Reports.
[Screenshot: The Review screen with the Report Title, Report Language, Source and Required Query]
When everything looks right, click Create Report! to start the analysis. Processing time depends on the size of your dataset. When the Report is ready, the wizard says "Here we go!" and View Report opens it. You can also find the Report on the Reports page at any time.
What You Get
Every piece of feedback in the Report carries the themes the Automatic Classifier found for it, each with its sentiment. The themes show up across the Report. The Overview lists them in the CONTENT CLASSIFICATION card and, with a sentiment classifier, as the best and worst performing themes. Content lets you filter records by theme. Performance ranks them by Satisfaction Score and Dissatisfaction Score, and Tables lets you cross them with columns and Dimensions.

To change how records are labeled, see Manage Labels in Reports. To reuse the themes on other data, open the Settings of the Report and click Create a Custom AI Model, which builds a model from the classified content of the Report. See Create a Custom Model with a Report.