Message Threads

Updated by Cassidy Chassagne

Claim and Release Texts

If you are running a texting program that requires your staff/volunteers to reply back to messages, you can have these individuals do this at their leisure using the “Claim Contacts” feature in the Switchboard Message Threads. Below we will explain how best to go about managing the claiming and releasing of your contacts.

  1. To see replies from recipients of your broadcast, head to your message threads from your Switchboard Homepage.
  2. This will bring you to the inbox dashboard. Claim contacts by clicking Claim Contacts under the Unclaimed column. This will allow you to claim batches of up to 10 contacts in need of a reply.
    1. Switchboard understands that campaigns want to build ongoing relationships with their supporters, so once a contact is claimed by any inbox user or admin, that contact will remain the responsibility of that user. Anytime the contact replies, whether its immediately or in a week, whoever claimed the contact will be responsible for replying.
    💡 This is a great feature for campaigns with volunteers who want to build lasting relationships overtime, run a relational program, or even maintain a consistent point of contact with the campaign for the voters they are texting.
  1. When you’re ready to reply, click “X Replies Needed” under Claimed by Me which will take you to an inbox where you can begin replying to your claimed contacts. If you have replied to everyone, and you still have time to spare, you can claim more contacts until there are 0 replies needed.
The “Claim Contacts” button claims 10 contacts at a time, so click the button as many times as you have time for!
  1. As mentioned above, these contacts will be auto-assigned to you across future broadcasts unless you release the contacts. In order to ensure responses are prompt, once a user is done with their shift or logging off, they can choose to release any claimed contacts so another user can claim them in the future! That way, even if the recipient’s response is delayed, they will always get a prompt answer from your organization. To release the contacts, click Release Contacts under the Claimed by Me column.
Admin: Allowing Users to Claim Contacts

“Message Threads” is your organization’s inbox. Anytime you send out a broadcast, responses will appear here. Anyone who needs a response from your organization for any broadcast will pop up at the top of your inbox.

In Message Threads, you can manage inbox users such as your teammates, staff, volunteers, etc. You can add labels to better organize contacts, and build lasting relationships with supporters, donors, and voters.

Switchboard keeps a record of the entire conversation with a recipient. You can see every response you have received from a recipient regardless of the broadcast. This allows organizations to have meaningful text conversations and drive supporter and donor retention.

This resource will teach you how to allow users to claim contacts:

  1. After you send out your broadcast, and recipients of your text start replying, head to your message threads from your Switchboard Homepage.
  2. As an administrator, this is where you will turn on the ability for volunteers to claim replies from certain broadcasts. Just toggle this switch to green for any broadcasts you want your inbox users to be able to access and claim messages from under Show to Users.
  3. You will then be able to see how many replies are unclaimed under Unclaimed. Users will be able to claim contacts in 10 reply batches.
  4. As an admin, you can track the process of claiming, replying, and releasing by inbox users in your account by clicking on Users With Claimed Contacts.
  1. This will open a new page where you can see everyone that has claimed any contacts within a broadcast. Of those, you can see how many still need a reply.
    1. If you think one of your inbox users is no longer active and forgot to release their contacts, you can always release contacts for them. This will allow unread messages to go back into the unclaimed pool of contacts for another inbox user to claim.
View All Threads: Your Switchboard Inbox
  1. After you send out your broadcast, and recipients of your text start replying, head to your Message Threads from your Switchboard Homepage.
  2. Click View All Threads to view your inbox. As an administrator Switchboard recommends this option so you can see very reply that has ever come into your Switchboard inbox from any broadcast you have ever sent.
Switchboard recommends this route to replying to inbox messages if you are a team of one or two. For larger teams of volunteers, we recommend the claiming and releasing structure within the Inbox Dashboard.
  1. This will open a new page where you will be able to reply directly to contacts.
    1. Switchboard’s inbox is unique in that we give users the full historical overview of a conversation with a specific contact. This allows users to understand the relationship the contact has had with the campaign, so they can respond with that context in mind. You can see the broadcast names and associated time stamps.
    2. You can reply by typing into the white box or using canned responses.
    1. You can see which labels have been applied to a contact to better understand the individual you are talking to, add an existing label, or even create a brand new label by typing into the box. This is great for keeping track of any important data that pertains to a contact.
    2. Switchboard Automatically opts out typical opt out phrases like “STOP” and “Stop2Quit” — however, you can always manually opt out individuals.
    3. Once you are done with a conversation, and you don’t want it to show up as a reply needed in the dashboard, you can click dismiss. If for any reason that contact replies again in the future it will pop up as reply needed again for whomever claimed the contact!
View All Threads: Edit Filters
  1. The Switchboard Inbox by default displays texts that need a reply. If you would like to view your texts differently, click Edit Filters.
  1. This will open a side bar on the right side of your page. You can change your filters to show contacts based on:
    1. Whether or not they need a reply
    2. Whether or not they have ever replied
    3. Their opt out status
    4. The last incoming message received time
    5. By Broadcast
    6. By inbox user texting them
    7. By label
    8. By result ordering
      1. Phone Number (ascending or descending)
      2. Last Incoming Message (ascending or descending)

Applying Labels to Inbox
  1. In Messaging Threads > View All Threads, click on the contact you want to reply to.
  2. Click Edit Labels above the text box.
  3. You will see all the current labels assigned to that contact. You can click the X next to the label to remove that label from your contact.
This will not remove a label’s action in NGPVAN if you applied this more than 10 minutes before you attempted to delete. Switchboard cannot remove label actions in NGPVAN that have been previously applied.

  1. If you want to add a label, click Add label.
  1. Here, you will see a list of all your labels. Choose the label(s) you would like to add to this contact. You also can search the label name in the search box at the top of the window. Additionally, you can manually create a new label by typing it in here (Note - you will need to go to the Labels section of the homepage if you want this label to have an NGPVAN action connected to it.)
Opt Out Scores

What is an Opt-Out Score?

Every day users use our platform to send and receive messages, but sometimes sifting through the incoming messages can be tough and time consuming. In response, we have developed a model that assigns labels to incoming messages based on the content to help sort your inbox and surface the most important replies.

To do this, we have trained a language model that classifies messages as they’re received and assigns likelihood scores. Currently the model rates the likelihood that a given response is an opt-out attempt, even if the respondent doesn’t use an opt-out keyword. Each message is scored on a scale from 0 to 1, with one being the highest confidence.

How can I use this?

You can find this in your inbox. Instead of exporting your incoming messages and searching for keywords in a spreadsheet, you can utilize our scoring system to filter the incoming messages right in Switchboard!

Open up the “Edit Filters” button, and scroll down on the menu till you find our Opt-Out scores:

The inbox will only show messages that are labeled with the scores you indicate. You can see exactly your range of scores above the inbox, for example:

About Our Model:

Like any model, the classification this model performs will not be perfect. Despite being accurate on average, it will make mistakes. Please feel free to ask questions about scores and we will monitor the model’s performance over time.

Along with acute problems with modeling, some general limitations also apply to language models like this one:

  1. The model may struggle with understanding the context and nuances of certain words or phrases, especially those with multiple meanings or those used sarcastically.
  2. It may not accurately classify text that contains spelling or grammatical errors, or unexpected sentence structures.
  3. Finally, since it was trained primary on english text, it may not handle languages other than the one it was trained on.

Performance

During our model training, we perform validation of the model’s performance. Scores can give a general sense of the quality of a model but are not conclusive because they are summaries of the model over all.

Those scores, and brief explanations, can be found here:

  • The model had an evaluation loss of 0.2, which indicates the model's error rate. A lower value suggests a better model.
  • We use the F1 Score of a model to represent the accuracy of the model in terms of both precision and recall. The model scored 0.8 where 1 is the best possible score.
  • The model scores 0.9 on the AUC ROC.This is a measure of the model's ability to distinguish between classes, with 1 indicating perfect classification.
  • Finally, the model correctly predicted the outcome 85% of the time in our testing data.

Training Details

To train our model, we first start with Google’s open-source BERT model trained on Wikipedia data and a corpus of book transcripts. We then fine-tuned this model for our specific use-case by training it on hand-coded examples of messages that clients might receive.

Citation

Devlin, J., Chang, M.-W., Lee, K., & Toutanova, K. (2018). BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding. CoRR, abs/1810.04805. Retrieved from http://arxiv.org/abs/1810.0480


How did we do?