.png)
Last updated: 6 October 2026
TL;DR
- Sentiment analysis of reviews judges for each review text whether the customer writes positively, neutrally or negatively and about what, so you find complaints the star rating does not show.
- Stars and sentiment often diverge: in car retail, customers sometimes give five stars with a complaint in the text so the dealer is not penalised by the manufacturer or importer (the brand).
- A top 5 complaints per location, with the trend versus the previous period, is the most useful format for dealer groups, because every complaint then gets an owner and a process.
- In Repmanager the dashboard shows sentiment and the top complaint themes per location with their trend, and in Reviews you filter on sentiment next to the star rating.
What is sentiment analysis of reviews?
Sentiment analysis of reviews is the automatic assessment of the tone and topic of a review text: is the customer positive, neutral or negative, and is it about waiting time, communication, the invoice or something else? Loose text becomes a countable overview of what customers complain about and what they are happy with.
The difference with a star average is that sentiment looks at the words, not the number. A 4.4 tells you things are going reasonably well. Sentiment tells you half the complaints are about call-backs. You cannot act on the first. You can act on the second.
Why are star ratings not enough?
Because the stars and the text often say different things. In car retail, dealer groups see customers giving five stars with a complaint in the text. The customer knows the dealer is judged on the rating and does not want to punish them. What that does to a KPI on rating alone is covered in chasing a 4.7.
The reverse happens too: a three-star review that is actually positive, from a customer who never gives five stars. Filter on stars alone and you miss complaints hidden in high ratings and see problems in neutral reviews that are not there. That middle group also often goes unanswered, as the forgotten middle shows.
Your own internal figure often drifts as well. Why an internal survey shows something different from Google is covered in internal score vs Google rating.
How do you get from review text to a top 5 complaints per location?
You have the text sorted into complaint themes automatically, count per location which themes come up most, and compare that with the previous period. The five biggest complaints with their trend are your agenda. Not for the whole group, but per location, because a workshop with a call-back problem gains nothing from a group average.
Dealer groups working with Repmanager, mostly in the Netherlands, that do this well follow four steps:
- Pick a fixed period, for example the past month or quarter, and always look per location.
- Take the five biggest complaint themes and check the trend: is a theme growing, or stable?
- Read five real reviews per theme. The theme tells you what it is about, the text tells you what exactly goes wrong.
- Give every theme an owner who can change the process, and check next period whether the trend turns.
The result is one overview per location you can show in the management meeting. Marketing managers describe it as the one picture per location they copy into their slides.
What sits behind a complaint theme?
A complaint theme is a symptom, not a cause. "Hard to reach" as a one-star review in the workshop is usually not about the phone system, but about a call-back request that was left lying. Knowing that, you fix it with a call-back routine, not an extra phone line.
| Complaint theme | What often sits behind it | Who fixes it |
|---|---|---|
| Communication | no call-back, no update on the car | reception and service advisor |
| Waiting time and speed | car ready later than promised, long wait at the desk | workshop planning |
| Appointments and planning | appointment forgotten or moved, no confirmation | planning and reception |
| Invoicing and payment | invoice higher than the quote, unclear items | service advisor and admin |
| Problem solving | issue not fixed after the first repair | workshop manager |
Complaint themes in car retail are mostly about the process around the car, not the car itself. How large that gap between sales and workshop is across the market is shown by the research on the aftersales leak.
Why do complaints rise in busy months?
Because there are more customers, not necessarily worse service. During tyre season and the busy inspection months more reviews come in, and so more complaints in absolute numbers. Count only the number of negative reviews and you get alarmed by a peak that is mostly volume.
Look at the share of negative sentiment instead of the count, and compare with the same period last year. If the share of one theme grows during the rush, that is a real signal: the process cannot keep up with the peak.
How do you set this up in Repmanager?
Repmanager analyses every incoming review for sentiment, topic and urgency. You turn that into a top 5 per location with the dashboard and the filters in Reviews.
- Check the industry in your Organization settings. It determines which topics the AI recognises in your reviews; for dealers that is Automotive.
- On the dashboard, choose one location in the organisation or location picker and a period, for example Past 12 months.
- Read the Sentiment block (positive, negative, neutral, with percentage and count) and the blocks ranking complaint and compliment themes, each with its trend versus the previous period and what it is mostly about.
- In the reviews overview, open More filters and combine negative sentiment with a high rating. That surfaces five-star reviews containing a complaint. Keep the view with Save.
- Filter on the intent "problem" or on high urgency to find complaints that need follow-up. Above each review an AI summary states the sentiment, the urgency and the topics.
- Build reply rules on sentiment, not stars alone, for example with the conditions sentiment equals positive and urgency is not high. Then a five-star review with a complaint does not get an automatic thank-you.
Which other reviews to answer automatically is covered in which reviews to answer automatically.
What usually goes wrong with sentiment analysis?
Filtering on stars only. You miss the complaints in high ratings. Always combine stars and sentiment.
Looking at group level. A group overview dilutes the problem of one location. Build the top 5 per location.
A top 5 without an owner. A list nobody picks up changes nothing, and neither does a coordinator who does not open the overview for two months. Schedule a fixed moment and assign someone per theme. How dealer groups divide those roles is covered in who replies to which review.
Ignoring compliments. The top compliments tell you which people and which steps customers value. Pass that on to the team.
Frequently asked questions
How reliable is automatic sentiment analysis?
Good enough to see patterns, not to follow blindly on a single review. Always read a few real reviews per theme. Besides positive, neutral and negative, the sentiment filter also has an undetermined option.
What do you do with a five-star review that contains a complaint?
Treat it as a complaint. Have a person reply to the point the customer raises, and pass it on internally to the location. How to build that reply is covered in how to respond to negative reviews.
How often should you review the top 5 complaints?
Monthly per location is the right rhythm for most dealer groups, with a quarterly comparison for the trend. Look weekly and you mostly react to noise.
Does sentiment belong in reporting to the manufacturer?
Usually not as a KPI, but as the explanation. The manufacturer judges on response time, response rate and rating; see how manufacturers hold dealers to account on reviews. Sentiment shows why the rating moves.
Ready to see your top 5?
Want to see which complaints top the list per location? In a 15-minute demo we show it with your own reviews.
