AI & Analysis
Read every comment, in every language.
Every comment, ticket and review gets a sentiment score, its emotions and a translation into your working language. Topics & Subtopics gives each one a place in the themes your customers raise. Your CX team sees how customers feel in every market, on the model your IT team chose.
Pick a time right in the form.
What did you like most?
Die Lieferung kam einen Tag früher als angekündigt, und alles war perfekt verpackt. Super!
- Language
- de
- Country
- DE
- Collected
- Oct 26, 14:32
What did you like most?
Translated to EnglishThe delivery arrived a day earlier than announced, and everything was perfectly packed. Great!
Original (de): Die Lieferung kam einen Tag früher als angekündigt, und alles war perfekt verpackt. Super!
- Sentiment
- Very positive0.86
- Emotions
- JoySurprise
- Topic
- DeliveryDelivery timeArrived early
Sentiment and emotions
Know how customers feel, not just what they scored.
Every comment gets one of six sentiment labels, from very negative to very positive, with Mixed for comments that hold both, and a score from -1 to 1 you can average and trend.
Emotions go further: joy, surprise, sadness, anger, fear and disgust. A comment can carry several or none, so you can tell a delighted customer from a merely satisfied one and know who to thank first.
MixedBoth positive and negative signals, scored near 0
- Arrived a day early and perfectly packed. Great!Very positive0.86
- Support sorted out my return within the hour.Positive0.54
- Lovely quality, though checkout took two tries.Mixed0.06
Arrived a day early and perfectly packed. Great!
JoySurprise
Please add an invoice address field.
No emotion, so no tag
Translation
One working language. The original stays.
Every open answer is translated into the language your organization works in and stored next to what the customer wrote. Teams in every market read the same comments, and nobody loses a nuance, because the original is one click away.
Sentiment, emotions and topics work across languages, so a German review and a French ticket land in the same chart.
30 languages → English
A global manufacturer brings feedback in 30 languages into English, tagged by topic, before it flows into its data lake.
- deThe delivery arrived a day earlier than announced.Translated to English
- frCustomer service replied in ten minutes. Thank you!Translated to English
- esThe app is very easy to use and checkout was quick.Translated to English
- nlBeautiful quality, exactly like in the photo.Translated to English
- itThe return was really simple, thank you.Translated to English
- plFast delivery and great contact with support.Translated to English
Topics & Subtopics
See what people talk about most.
AI reads the open text in a dataset and builds up to four levels of topics and subtopics, named for you. Each comment is placed in the theme it belongs to, whatever language it was written in.
Open a subtopic to read the comments behind it, with their sentiment and the original language a click away. You see that Arrived early is 712 comments at a glance.
- Delivery1,640
- Delivery time1,210
- Arrived early712
- On schedule498
- Packaging430
- Checkout1,190
- Support1,034
- Product quality948
Delivery timeArrived early712
The delivery arrived a day earlier than announced.
Post-purchase surveyVery positiveTranslated to English
Parcel came early, so I could gift it on time. Thanks!
Support ticketVery positive
Customer was happy the order beat the estimate by two days.
Call notePositiveTranslated to English
Reporting
Put feelings on the dashboard.
Group any chart by sentiment, trend the average score, or filter to the comments that carry joy. Share the dashboard and everyone with read access sees the same numbers.
For IT, the enrichment is plain typed columns next to every record in Formbricks Hub, in your own PostgreSQL. Power BI, Superset and your data lake read it with SQL, with nothing extra to build.
Comments by sentiment
- Very positive1,204
- Positive1,786
- Neutral842
- Mixed391
- Negative446
- Very negative143
All comments4,812
Average sentiment score
By month
Comments with joy
Emotions contains Joy
- Survey1,104
- Support386
- Calls92
The same fields as columns in your own database, for SQL and BI tools
Your model
On the model your IT team already approved.
Point each enrichment at the provider and model you choose: an open model you host behind an OpenAI-compatible endpoint, or Google Gemini. App features such as AI survey translation and AI chart creation also run on Azure OpenAI and Amazon Bedrock.
Self-hosted on your own model, feedback stays inside your network. There's no new AI vendor and no new sub-processor to review, and your compliance team signs off on infrastructure it already knows.
Your infrastructure
- FormbricksSurveys, topics, dashboards
- Formbricks HubRecords and enrichment, on PostgreSQL
- Your model endpointAn open model you serve
- TranslationOpenAI-compatibleQwen on vLLM
- SentimentOpenAI-compatibleQwen on vLLM
- EmotionsOpenAI-compatibleQwen on vLLM
- SearchOpenAI-compatibleAn embeddings model you serve
A model for every job
Translation, sentiment, emotions and search each get their own provider and model. Use a small, fast model for one and a larger one for another.Nothing runs without a model
Enrichment starts only once you set a provider and model for it in Formbricks Hub.Sentiment and emotions, each on or off
Each has its own switch in the Hub's tenant settings, on by default. Turn one off and it stops for new and edited records.Backfill existing records
Switching enrichment on for feedback you already have? One backfill command classifies every record that's missing a result, and skips the ones already done.
Customer stories
Trusted by teams that own their data
See how public sector, research and product teams run feedback on Formbricks.
Frequently asked questions
Which license includes it?
Which feedback gets enriched?
What happens when a comment is edited?
Can we rename topics?
Where does it run?
- Read the code
- ISO 27001 certificate on request
- SOC 2 Type II report on request
- Accessibility report available
See your own feedback, enriched.
Bring a sample of comments in the languages your customers write. We'll add sentiment, emotions and topics to them on the model your IT team approved.








