How to Run AI Sentiment Analysis on New Reviews in Make.com

Reading every review is not scalable. This scenario sends each new review to AI for a sentiment score and a one line reason, then files positive and negative feedback into separate tabs so trends are easy to spot.

• A Make.com account
• A source of reviews (a webhook, form, or Google Sheet)
• An OpenAI API key
• A Google Sheet with Positive and Negative tabs

Step 1: Capture the review

Begin with the trigger that fits your source. A Webhooks Custom webhook works well if reviews come from your app. Run once with a sample review so Make maps the text field.

Step 2: Score sentiment with OpenAI

Add OpenAI Create a Completion (Chat). Ask for a strict label and a short reason in JSON. Forcing a fixed vocabulary (positive, neutral, negative) makes the next routing step reliable.

Classify the sentiment of the customer review. Reply with ONLY this JSON, no extra text:
{ "sentiment": "positive" | "neutral" | "negative", "reason": "short phrase" }

Step 3: Parse and route

Add a JSON Parse JSON module to turn the result into fields. Then add a Router with one route filtered to sentiment equals positive and another filtered to sentiment equals negative.

Step 4: Log to the right tab

On each route add a Google Sheets Add a Row module pointing at the matching tab. Write the review text, the sentiment, the reason, and a timestamp so you can chart volume over time later.

Step 5: Test with mixed input

Send three sample reviews of differing tone. Confirm each lands in the correct tab with a sensible reason. Then activate the scenario.

Result: a live, sorted record of customer sentiment that shows exactly where complaints cluster, with no manual reading.