Getting your data in: spreadsheets, fifty names for email, and everything German text hides
Large teams do not send feedback one message at a time. They export spreadsheets, they label the same channel a dozen different ways, and their messages are full of postcodes and case numbers nobody has time to tag. All three are now handled on the way in.
By Howzer Team, Product ·
Picture the actual situation. Somebody exports six months of messages from a service desk. The file is semicolon separated because it came out of a German spreadsheet, the encoding is whatever that machine used, the channel column contains eleven different spellings of email, and every third message mentions a postcode and a case number in the body text. Cleaning that up by hand used to be the first day of a rollout.
Upload the file as it is
The import endpoint works out the encoding and the separator by itself, matches your columns to the right fields, and reports clearly on anything it could not place. Every imported row then receives exactly the same full analysis as a message submitted live. No preparation scripts, no required format.
Fifty ways to say email
E-Mail, email, Email Eingang, Mailbox, elektronische Post. All one channel, and that is just the first of nine. Across German and English spellings we found more than fifty labels in real data. They are now recognised and folded together automatically, so a report by channel shows nine rows instead of fifty.
- Email, phone, chat and social media, each with the informal spellings people actually use.
- Web forms, letters and fax, app feedback, and survey responses.
- Anything unrecognised is kept as such and logged, so the list keeps growing from real traffic rather than guesswork.
What German text quietly contains
A German service message carries a surprising amount of structure in plain prose. Howzer now reads it out on the way in, without anyone tagging anything.
- A five digit postcode becomes a city, a federal state and a pair of coordinates, which is what makes regional patterns visible at all.
- Case numbers, ticket numbers, contract and order references are pulled out of the running text.
- Words like dringend, sofort or schnellstmöglich are recognised as the urgency signal they are.
And when the context arrives later
Not everything is known at the moment a message lands. A separate endpoint lets you attach a provider, a case reference, tags, notes or a manual priority to a message that was analysed days ago, which is what you need when your own systems catch up afterwards or when somebody wants to annotate a case by hand.