Atlanta Scribes

Learn Our Process

Want to know more?

We aim to make our process open for others to learn so they can implement and improve on our ideas. Just contact us with 'Learn More' in the 'What does this message relate to?' field and we'll get back to you as soon as possible. We are happy to help.

The Problem

Call volume
The Altanta City Council was receiving a high volume of calls, which was overwhelming the capacity of the city’s servers and IT Team.

Technical Limitations
The City’s current voicemail system can only hold a certain volume of voicemails, and the IT Team had to be available to move calls off one server and onto another, in order to make room for additional voicemails.

Voicemail Metadata
The current voicemail system does not prompt the caller to identify their name, location /residence, their stance on the issue, etc. This creates a lack of consistency in the resulting data, and creates analysis problems.

Call volume and listening / analysis
The Atlanta City Council by law has to listen to every public comment. The volume of calls in itself created hours of listening time for the Council. However, technical hiccups with the current process for storing and replaying the messages made those listening sessions even longer.

Additionally, the Code for Atlanta Team had to devise a process whereby they could listen / transcribe / and analyze this data, and then turn around the information to the Council in an effective time frame.

Receipts
If a resident called and received a ‘voicemail inbox full’ message, it created frustrations and fears that their public comment would not be received.

 

Existing Process

Atlanta City Council:
1. Cisco powered phone lines are run by City Hall staff and collect voicemails from the public.
2. Voicemails are played at the city council meetings.

Code for Atlanta:
3. Voicemails are shared with volunteers at Code for Atlanta.
4. Voicemails are transcribed via Otter.ai
5. Transcripts are put into a google spreadsheet and volunteers extract information including:
a. Fix errors in transcription.
b. Extract additional information:
i. Name
ii. Location
iii. Overall sentiment
6. Final spreadsheet is exported to Qutee.com’s system for analysis.
7. Reports are generated that include full transcripts and summary information.
8. Reports are shared with council staff and members.

Goals for Improvement

A. Fix technical limitations
B. Speed the process
C. Reduce reliance on volunteers
D. Collect metadata directly from callers
E. Provide additional channels
F. Share with the public
G. Automate the system so the council can apply this to all legislation
H. Share the system with other cities and government institutions

Future Process/ In Development

Web app for volunteer interface

Promotion Website

Twillio Collection