A 'verbatim' is what we call a single unit of freeform text communication; such as an email, a survey response, a note, or a chat or phone call transcript. Verbatims are grouped together in sources.
Below is an example of how a verbatim is presented in the Explore page of the user interface.
An example email sent to an insurance underwriting mailbox
Verbatim metadata
Every verbatim has associated metadata that consists of structured data points that provide additional information about the communication or conversation and its participants.
All verbatims are required to have an associated timestamp — which typically corresponds to the time at which that verbatim was originally created.
In addition to timestamps, the platform typically stores additional metadata associated with each verbatim. Typical examples of metadata fields are:
- Name and contact details of the message participants.
- Sender and receiver domains for emails
- Number of messages in a chat conversation or number of emails in a thread
- Demographical data, such as gender, age, country etc
- A quantitative measure of customer satisfaction with the interaction, such as Net promoter score (NPS), Customer Satisfaction Score (CSAT) etc
- For phone calls, the raw audio used by the platform to transcribe it
- 3rd party IDs when the Verbatims are imported from another system, for example the email message ID from an Exchange server
Example metadata associated with a verbatim
Some datasets may contain verbatims with PII (personally identifiable information) in their metadata (e.g. `userId`), these fields can be marked as sensitive. By marking these fields as sensitive to view this metadata requires enhanced user permissions.
Having rich verbatim metadata allows users greater ability to train and analyse their data within the platform, as users can filter by metadata fields within both Explore and Reports.
Previous: Validation