Triage dealership customer texts by sentiment
A car dealership group's Make flow reads customer texts, closes routine ones, and flags unhappy customers.
What they did
Every inbound text runs through a sentiment check. Routine messages get closed with a reply; negative ones go straight to the aftersales team.
How it went
It auto-closed about 18,000 of 47,000 messages, so staff only saw the ones that needed a person.
Our read
Our analysis · Oct 28, 2026Not from the source. Our own assessment of the approach, which may date as the tools change.
Where it falls down
- Sentiment is a crude proxy for urgency: a calmly worded message about a safety fault reads as routine.
- 18,000 of 47,000 auto-closed is a volume figure, not evidence that none of those customers needed a person.
- Short replies and sarcasm are exactly where sentiment scoring is least reliable.
Worth trying next
- Sample closed conversations weekly and measure how many should have been escalated.
- Route on safety and legal keywords regardless of what the sentiment score says.
Try it by SolvedWith.ai
A starting prompt we built from this case. Not the original poster's wording.
I work in retail & e-commerce and want to triage dealership customer texts by sentiment. Walk me through setting this up with Make, step by step. Before anything that sends messages as you, stop and ask me first.