AI Theme & Sentiment AnalysisUnderstand what customers praise and what needs attention

Go beyond star ratings. Cacao uses AI to classify feedback into your categories and subcategories and identify praise and complaints within each theme—even when the same review contains both.

7 days free. All Cacao features. No credit card required.

Theme & Sentiment Tracking: one review, several findings

A star rating summarizes a visit, but the written review can describe several different experiences. Cacao can assign multiple analytical categories to one review and identify different sentiments within those themes.

Illustrative restaurant review: “The burger was excellent, but we waited far too long.” The food receives praise while the wait time receives a complaint. Both findings come from the same review.

Illustrative classification of one review; categories are configured by the business.
Review excerptCategory → subcategorySentiment
“The burger was excellent”Dishes → BurgersPraise
“We waited far too long”Service → Wait timeComplaint
  • Find complaints in highly rated reviews as well as praise in otherwise negative feedback.
  • Keep the sentiment attached to the relevant theme instead of treating every part of a review alike.
  • Read the original comment before drawing a conclusion.
Cacao review dashboard, translated into English and anonymized. Worst ranked is selected within the category.

Cacao reviews within the Product category, with category cards, account and location filters, Worst ranked selected, and additional categories beside each review

Scroll to explore the enlarged image.

Cacao reviews within the Product category, with category cards, account and location filters, Worst ranked selected, and additional categories beside each review

Explore reviews by category, from worst to best

Select a category to read its reviews and investigate the original comments. Use the Worst ranked option to start with the feedback that needs attention, then work through to the best-ranked feedback.

Search within the category and filter by account, location, and date. Review the rating and additional categories alongside each comment: the same review can belong to more than one analytical category.

Category configuration in Cacao. This English-language screenshot shows category descriptions, status, and example alert recipients.

Cacao category configuration listing Product, Operations, Delivery/Apps, Systems, Marketing, and Uncategorized, with descriptions, status, and alert settings

Scroll to explore the enlarged image.

Cacao category configuration listing Product, Operations, Delivery/Apps, Systems, Marketing, and Uncategorized, with descriptions, status, and alert settings

Build categories and subcategories around your business

Create and edit the categories your team needs. A restaurant can use Dishes as a category and Burgers as a subcategory, alongside topics such as service, wait time, cleanliness, and delivery. These are examples, not a fixed taxonomy.

Use the same category definitions across locations so your teams can investigate comparable feedback. Correct classification and sentiment when a result needs review. Editing a category does not automatically reclassify historical reviews.

  • See your categories, their descriptions, and whether each category is active.
  • Open a category to edit its definition and associated alert settings.
Editing the Product category in Cacao. The category definition describes the topic; alert settings appear separately below it. Recipient addresses shown are examples.

Cacao Product category editor showing name, topic description, guidance to include positive and negative feedback, active status, and email alert settings

Scroll to explore the enlarged image.

Cacao Product category editor showing name, topic description, guidance to include positive and negative feedback, active status, and email alert settings

Define the topic your AI category should recognize

Edit the category name and description to explain the feedback it covers. Describe the topic itself so the category can include both praise and complaints—for example, taste and portion size, rather than only bad taste or small portions.

The editor also lets you manage whether the category is active for future AI assignments and configure its email alerts. Defining an analytical topic and choosing how to notify the responsible team are distinct settings.

Compare themes across locations, groups, and periods

Filter category results by location, customer-defined group, and date window. Inspect praise and complaints to understand whether an issue is concentrated at one branch or appears across several locations. Select another period to investigate how the feedback changes.

A review counts once within each applicable analytical category and may appear in several categories. The matrix uses all written comments at the location in the selected scope as its denominator, so category percentages do not have to add up to 100%. Read the comment counts alongside the percentages.

Analyze Google reviews and Cacao internal survey feedback in separate views. They remain distinct sources rather than a combined rating or comment dataset.

Explore the Google reviews dashboard

Use customer feedback to investigate operational changes

Use recurring themes and original comments to decide what deserves investigation. Restaurant teams can review recipes, service practices, staff performance, or delivery policies. The team evaluates the evidence and makes the operational decision.

For example, repeated comments about burgers can prompt a recipe review; waiting-time complaints can prompt a review of service practices; delivery complaints can prompt a review of packaging or shipping policies. These are illustrative uses, not attributed customer outcomes.

1. Identify the theme

Find praise or complaints within the categories relevant to your business.

2. Investigate the comments

Read the original feedback and compare locations and date windows before deciding what it means.

3. Agree the operational response

Your team decides whether to adjust a recipe, change a service practice, review staff performance, or revise a delivery policy.

4. Review subsequent feedback

Inspect later comments in the same scope. A change in feedback alone does not establish what caused it.

Connect analysis with the right follow-up workflow

Analytical categories can be multiple per review. Negative-review alerts use one routing category and configured recipients to notify the responsible person. Theme analysis and alert routing serve different purposes. Identifying a complaint and triggering an alert does not stop response publication in Automatic mode. Choose Hybrid mode, which automatically answers only four- and five-star ratings without comments, or Semi-automatic mode when written feedback must be approved before submission.

This feature focuses on understanding themes and sentiment. For the wider workflow of metrics, location reporting, Excel exports, and qualitative summaries, explore review analytics software for multi-location businesses.

Understand the themes behind your customer reviews

Bring categories, sentiment, and original comments together so your team can investigate what matters at each location.

Get started

Frequently asked questions

Can one review have several themes and different sentiments?

Yes. One review can praise a burger and criticize the wait time. Cacao can classify the review into multiple analytical categories and identify praise and complaints within the relevant themes.

Can we customize review categories and subcategories?

Yes. Your business can create and edit categories and subcategories, such as Dishes → Burgers. Users can correct classification and sentiment. Changing a category does not automatically reclassify historical reviews.

Is sentiment analysis based only on the star rating?

No. Written feedback can contain complaints alongside a high rating or praise alongside a low rating. Cacao analyzes the themes in the text so teams can inspect those findings separately from the overall rating.

Can we compare feedback across restaurants and periods?

Yes. Filter category results by location, group, and date window, then inspect the original comments and counts. Google reviews and internal survey feedback remain separate analytical sources.

Does Cacao automatically make operational or staffing decisions?

No. The analysis helps your team investigate customer feedback. People decide what action to take, and operational implementation and follow-up remain the organization’s responsibility.