Sample AI Sentiment Analysis Report.
A full sample output based on a plausible scenario — 500 reviews of a fictional robot vacuum, with aspect-based sentiment scoring, confidence bands, and grounded citation excerpts.
What this sample shows
Aspect-based sentiment for a fictional HomeSweep Robot Vacuum 500 across 500 reviews: overall polarity distribution (58% positive, 22% neutral, 20% negative), six aspect-level scores with confidence bands ranging from −44 to +72, and a citation excerpt per aspect showing the specific review passages behind each score. Deterministic grounding means every score is traceable to source text.
HomeSweep Robot Vacuum 500
500 reviews · 6 aspects detected · grounded with 18 citations
Polarity distribution
Aspect-based sentiment
Grounded citations — one review passage per aspect
"It actually picks up the dog hair that my old vacuum smeared around. First pass on a rug and you can see the difference. Suction holds on max mode for the whole cycle."
"The map was accurate after the second run. It found the alcove behind the couch and named the rooms reasonably. Lidar version, not just bump navigation."
"Quieter than I expected on carpet. Hard floors it picks up a whine but nothing that stops a conversation in the next room."
"Advertised 90 minutes, I get about 65 once it's been charged through a few cycles. Enough for one floor, not two."
"The app drops the unit every few days. I have to re-pair and then reset the schedule. The schedule is the whole reason I bought this."
"Support never responded about the dock light behavior. The online chat timed out twice and the email took eight days for a template reply."
How aspect-based sentiment works.
Aspect-based sentiment analysis (ABSA) identifies the features, services, or experiences mentioned in text and scores polarity per aspect rather than per review. A review saying "great suction but the app is terrible" produces two aspect scores, not one overall score. For product teams, this is the more useful view because it identifies which specific aspect to improve.
Indellia's ABSA implementation grounds every aspect score in retrieved citations from the source corpus. Rather than an LLM producing a number from opaque reasoning, the system finds the passages that justify the score and returns them with the output. Confidence is a function of citation strength and semantic consistency. See the ABSA glossary entry and deterministic AI.
How Indellia built this sample.
The sample is a fabricated scenario chosen to illustrate realistic aspect-based output. The product (HomeSweep Robot Vacuum 500) is fictional, and the review excerpts are written to match the kind of language real reviews in this category use — specific, mixed, and anchored to named aspects. No real customer data was used.
The engine producing the aspect-level scores and confidence bands is the same one running on the live Indellia platform. When you submit text through the AI Sentiment Analysis Tool, the output matches this format. For aspect-based sentiment explained in depth, see sentiment analysis for product reviews.
The confidence band is tied to two factors: citation strength (how many source passages support the score) and semantic consistency (how much the passages agree). Low-confidence scores (below 0.6) are flagged and should be treated as hypotheses until more text arrives.
Keep going.
Have a specific question?
Indellia's AI agents answer with citations from real customer feedback across Amazon, Walmart, Best Buy, and 20+ retail channels.
Aspect-based sentiment, grounded in your own citations.
Continuous ABSA across every review, ticket, and survey response — per SKU, per aspect, per retailer.