Texting team deep dive: problem-solving in scaling a non-traditional voter contact program

By: Eric Babb and Autumn Frlekin

The NextGen Distributed Texting Team encountered many of the same successes and barriers that others had in 2020. As with any distributed texting program, there is the fundamental issue of matching the number of texts available to send, the number of volunteers taking time out of their day to send them, and the number of volunteer leaders who are trained to run the show. We had a large number of extremely well-trained volunteer leaders on top of inquisitive part-time fellows who took the initiative to try new things and develop a positive, cyclical volunteer experience that provided feedback to almost every volunteer.

Staffing structure and Volunteer Leaders

Everything that follows below — our incredible sweeping program, our fast and always effective campaign moderating, our awesome training and onboarding programs, and our clear campaign day announcements for the hundreds of volunteers texting with us each day — would not have been possible without our 67 volunteer leaders and six fellows. When we first started texting for the very real National Vote By Mail Day holiday in July, we had a small original group of just five dedicated volunteer leaders whose sole focus was adding incoming volunteers (manually) to campaigns and moderating volunteer questions. As the volunteer leader team grew from five to around 30, we began to rely more on volunteer leaders as captains of different teams. One volunteer leader became the captain of our moderators, another became the shift captain who would help with launching campaigns and making sure all volunteers who wanted texts could join, and another helped get our sweeping program off the ground.

Our ability to work with our volunteer leaders changed dramatically when we brought on six fellows, four of whom had been volunteer leaders. With our staff team of two staff and six fellows, our ability to check in with all volunteer leaders, ask them what they enjoy, get their feedback, and make programmatic improvements was greatly increased. Hiring volunteers who already knew the program so well as fellows and giving them the agency to make quick improvements resulted in a program that could easily manage all of our volunteer leaders, texters, and program components.

As we moved into GOTV, we asked our volunteer leaders to join a single team (or two). Led by a fellow or volunteer captain, these teams would work the same hours during GOTV in an effort to build team camaraderie and efficiency. Because of the work put in by fellows prior to GOTV to document all processes, volunteer leaders were able to step into a GOTV shift that functioned in a very similar fashion to any other shift during the election.

So what did our fellows and volunteer leaders help us accomplish? We’re excited to show you each component of our program. We will begin with our quality control component, called sweeping, since it was the piece that tied everything else together. Once our incredible fellow, Ellen Parkhurst, developed our sweeping program, there wasn’t a single other piece of our Distributed Texting program that wasn’t affected.

Sweeping

Any texting campaign will have three quality control goals: to keep out bad actors, data integrity, and to give your volunteers feedback on their texting. The process of quality control in texting is generally referred to as sweeping.

At first, our sweeping only focused on the first piece: looking out for bad actors. As we built capacity, we noticed that the feedback we gave volunteers was generally well-received and resulted in more engaged pre-shift huddles. The feedback might revolve around ensuring volunteers are collecting relevant data, correctly opting out voters, or appropriately engaging with voters. Overall, sweeping is one more chance to stay connected to our volunteers, so that they hear from someone personally instead of just clicking on a campaign link.

Our sweeping process begins with a campaign tracker Google spreadsheet that has two broad functions. The first tab is essentially the table of contents that includes a row for each campaign. The columns on this tab are: a link to the backend of the ThruText conversations for that campaign, cells for sweepers to indicate their progress, and a link to the campaign sweeping tab in the spreadsheet. You want to imagine a spreadsheet where one tab is the table of contents, and each successive tab is each individual campaign.

Organizations like NextGen will begin the process of sweeping by asking what we want volunteers to do well. Typically, that means opting out voters correctly, filling out survey questions, and replying appropriately to voters. After exporting ThruText campaign data, it only takes a bit of spreadsheet know-how to create a campaign sweeping tab in your tracker that includes the following: the name of every volunteer who texted in a specific campaign, the number of texts they sent, the percent of voters they opted out, and the percent of surveys they completed.

Once a campaign tab was ready to be swept, sweepers would self-assign volunteers whose conversations the sweeper would review. The campaign tab included a number of checkbox columns that could be checked for each voter a volunteer texted with if an issue was observed. We had checkboxes for: not opting out voters, opting out voters incorrectly, engaging inappropriately to voters, incorrect survey usage, and a couple of others. In a touch of sweeping magic, our sweeping fellow, Ellen, created a formatted cell next to each voter name that would indicate whether a voter had no issues, needed coaching, or should be put on hard stop because of a major issue. This made it very easy for staff to quickly review the sheet and see who to reach out to. And that is how sweeping closed the loop on our volunteer experience and made all of our texters the best texters they could be.

For volunteers who needed coaching, were put on hard stop, or were first-time volunteers and we wanted to check in on them, Ellen created a list of scripts that sweepers would send to each volunteer. Volunteers had two additional checkboxes next to their names to indicate that they had both received a direct message over Slack and that they had responded and any issue had been resolved. This is the piece that made our volunteers better texters because, as we would see in our pre-shift huddles, providing written feedback meant volunteers could come in with very specific questions. For example, a volunteer might come into a pre-shift huddle and ask, “on the last campaign I chose XX survey option, but was told I should have chosen YY instead. Can you explain this to me?” The trackers created by Ellen, the sweeping done by our sweepers, the feedback provided by our sweepers, all came together each campaign so that our volunteers were just a little bit better for the next one. Our sweeping program was the piece that made our volunteer experience a full loop, and caused a dramatic improvement in the quality of texts sent.

Our sweeping program was only possible because of the large number of well-trained volunteer leaders that we had on our distributed team and their dedication to provide constructive and positive feedback to all of our volunteers.

Pre-Shift Huddles

At the start of each campaign, we would take 15 minutes on Zoom to go over the goal of the campaign; the script, its initial message and recommended replies; the survey questions and data for volunteers to collect; and finally, we would provide feedback based on what our sweepers had observed while sweeping the most recently swept campaign. So for each campaign, we could reiterate certain messages we wanted our volunteers to hear, whether it was about opt outs, archiving messages, sharing links with candidate websites, etc.

Pre-shift huddles were also the only time outside of a training when volunteers would see the face of another volunteer. With our work being done over tools such as Slack and ThruText, it was important to maximize opportunities for face to face interaction, particularly during 2020 when so many of us were isolated more than usual.

Training

Training was another component that was positively affected by our sweeping program. As we began to notice what questions texters had when they received sweeping feedback, we would work to incorporate answers to those questions into our training. Anecdotally, our sweepers noticed over time that new volunteers had fewer issues with items such as survey responses and opt outs — areas of focus that we emphasized more in trainings after receiving feedback from sweepers.

Many volunteers who were new to texting were new to Slack as well. Therefore, the first half of each training was geared towards helping new volunteers understand how to use Slack: how to find and join a campaign, where to ask questions, where to find direct messages from Text Team Leaders, etc. The other 50% of a training would be focused on training new volunteers on how to use ThruText.

Prior to GOTV, we held trainings at various times throughout the week. During GOTV we decided to hold trainings at the start of each shift. Each training would start with a ‘pre-shift huddle’ for all texters, new and returning. After going through the script and surveys, returning texters would head over to our private texting channels and start texting. New texters would stick around for a fuller training before completing their onboarding and joining a campaign.

Like many organizations during 2020, we often had more volunteers than we had texts to send. In practice, this meant potentially running out of texts in a shift before new volunteers were able to complete their training and onboarding. The most effective way we found to handle this was a friendly post from a shift captain asking volunteers to only take one batch of texts at the start of a campaign. It was very much an honor system, but one that worked out for us. Another option we sometimes used was to ask that only volunteers who signed up for a texting shift on Mobilize join us for the day. With 20,000+ volunteers in our Slack workspace, it was easy for ‘walk in’ volunteers to take a large number of texts, leaving fewer for volunteers who did sign up for that shift.

New volunteers also had access to a written training guide that was available both inside our Slack workspace and at organizing.nextgenamerica.org. Making the training guide available outside of Slack was important for us so that volunteers could bookmark the website and not use Slack if it wasn’t their preference.

Onboarding

We valued onboarding our volunteers in a way that was easy and accessible. Upon joining Slack, volunteers would receive a Direct Message with initial steps to take, including steps on how to onboard.

The first step for volunteers to complete was to read through our Text Team Guide and take our Text Team Quiz. Once they received a passing score on their quiz, they would post in our #texting-onboarding channel to let our volunteer leaders, Greeters as they were known for this role, know they are ready for the next steps. A Greeter would then react to their post with a :heavy_check_mark: emoji and DM them with instructions on establishing a ThruText account with us. After creating their ThruText account, volunteers would notify their Greeter who then would check for their account and add the volunteers to the channels needed to text in a campaign with us, #text-links and #text-questions.

In order to make our text guide and quiz easy for volunteers to access, we established slackbot reminders to go off at the top of each hour with the on-boarding instructions along with links to our guide and quiz. We also had a fellow or staff member post in #texting-onboarding each morning to let our volunteers know when our Greeters were coming online to start onboarding and provide the guide and quiz link once again. This post would be “pinned” in the channel for the remainder of the day. We would post again towards the end of the day to let our volunteers know that onboarding was “closed” for the day and our leaders would no longer be online. One of our fellows created a “Hover Here” for each of our texting channels. She did this by changing the channel topic to “Hover Here :hand:”, when users placed their mouse over the “Hover Here” useful links for that channel were provided. In #texting-onboarding, the “Hover Here” had quick links to our texting guide and quiz and directed volunteers to the channel’s pinned post, another way we ensured these training materials were easy to access and visible to volunteers.

There were a lot of benefits to having such a personalized on-boarding process. Volunteers were able to use DMs to communicate any issues they were having to the greeter that was onboarding them. Our greeters were determined to give everyone a chance to text with us. Even if it took a few hours or even days to get a volunteer texting, our greeters wanted to make it happen! For many of our volunteers, our program was their first experience with Slack, so an online workspace was all new to them. Our greeters took the time to find the best way to troubleshoot issues with new volunteers and get them texting. Another benefit of the manual onboarding process was that we were able to gather feedback from our greeters about what steps of the onboarding process was confusing to volunteers and ways we could update our guide and quiz to improve upon those steps. In addition, if there were basic issues that our sweepers were seeing in a campaign and they felt it needed to be communicated to new texters, we could easily incorporate that information into our onboarding by updating our guide and greeter resources and having greeters remind new texters.

Scripts

As the election grew closer, we inserted creativity into our scripts and altered our language to appeal to the voters in our various universes. We knew that voters were being contacted by several different organizations and candidates and we wanted our messaging to stand out and increase engagement. Although we were already able to track the response rate for a given campaign through ThruText, our Sweepers were able to give us a more detailed idea of how voters reacted to our scripts. They would report if the response was overwhelmingly positive or negative, if there were questions that voters had that were not included in our recommended replies, if the language we used was confusing, etc. By GOTV, we had an expansive list of detailed recommended replies containing different links and resources as well as answers to common questions that were seen by our sweepers. One script format that had an increased response rate was giving voters a one number response to choose from. For example, an initial text would ask where a voter was in their voting journey and ask them to respond with the corresponding number, 1 — already voted, 2 — need to send in my ballot, 3 — need to register to vote, etc. These scripts had a fun tone and made it easy for voters to let us know their needs and, in turn, easy for our texters to provide them with the appropriate resources.

Links

There were a number of small issues that we resolved over time in regards to creating a process for volunteers to join a campaign, and we’ll detail those below. But nothing helps volunteers more than a super clear post that informs volunteers what they will be doing and when, where to go for resources, and provides helpful reminders. These posts, like other documentation, evolved over time based on feedback from our fellow, Olivia Ghiz.

Our #texting-links (formerly #text-requests) channel was where volunteers would go to access our ThruText link to join a campaign. During our GOTV dry runs and day to day texting, we experienced several volunteers sending messages such as “ready to text” “batch please” etc. as the campaign start time drew near. These messages would overwhelm the channel and make it difficult for other volunteers to access our campaign announcements. We realized that the name #text-requests could be misleading as it implied that volunteers needed to request their texts. Thus, we transitioned the channel name to #text-links as it better represented how the channel would be used. Our shift captain for the day would post an announcement that contained the ThruText Link and all volunteers had to do was click the link to join a campaign!

This name switch did help alleviate some of the posts that would overwhelm the channel but we were still seeing volunteers having issues accessing the campaign link and sending lots of posts. We took a look at the templates we were using for our campaign announcements and edited them to be shorter so it would be easier for volunteers to take the information from them and understand the process. We included information on which state we were texting, links to the script and any special instructions for the campaign along with the ThruText link. We also utilized emojis to make important information and links stand out. Again, we utilized the “Hover Here” topic and would update it to contain the campaign link when we had an active campaign and direct volunteers to the pinned posts. When we did not have an active campaign, we would provide the date and time of our next campaign along with the link to ThruText to check replies from past campaigns. The “Hover Here” became a quickway for volunteers to see where we were that day in regards to campaigns.

In addition to making information and resources available to volunteers, we were sure to respond to all of their posts in the links channel as well. If a volunteer’s post was better suited for another channel, we would reply in a thread directing them to the correct channel; if they were confused about when there would be texts, we would reply to them with the start time of the next campaign, etc. We wanted to ensure that each volunteer felt valued and supported by our team and by responding to each post, even if unrelated to the channel, we let volunteers know that they were being heard and that we were there ready to help.

There were times, however, when we would delete posts in order to stop a “spiral.” Until September, ThruText did not have an option for volunteers to join a campaign by clicking on a link; volunteers had to be manually added! Part of our previous process asked volunteers to type “ready to text” in our texting channels at the start of a shift so that moderators could add them to a campaign. Well, old habits die hard and we learned that if “ready to text” shows up once in the texting channel, especially on a busy day, it would only be a matter of moments before other volunteers would follow suit. If your goal is to keep your text campaign channel clear for other volunteers to find the information they need, deleting certain posts and including friendly language that certain posts will be deleted can go a long way in keeping the positive vibes of a channel while at the same time allowing the channel to remain helpful for all volunteers, new or returning.

Moderators

The most apparent impact of our Sweeping program was seen by our moderators. The role of our moderator volunteer leaders was to monitor our #text-questions channel and answer any questions text volunteers had during a campaign. Once our sweepers began to regularly contact texters with feedback about their texting, moderators saw a shift in the type of questions being posted in the #text-questions channel. Texters seemed to have gained a deeper understanding of our program and how our processes worked. The majority of their questions were around campaign specific information rather than the functionality of texting. The #text-questions channel was another channel that utilized “Hover Here” for the channel topic. “Hover Here” in #text-questions directed volunteers to pinned post, linked to the script and the ThruText replies link.

Regardless of what the question pertained to or how many times it was answered before, our moderators crafted a response to every question posted in the #text-questions channel. There were certain responses we automated by creating Slackbots for such as our opt out guide, voter guide and the voter protection hotline but even questions that had a Slackbot provided link, a moderator would still reply to the volunteer in a thread to make sure they received the answer they were looking for. We prided ourselves on keeping our response time under 3 minutes and our well prepared and well staffed moderator team kept under this 3 minute mark for the majority of GOTV. Our moderators were the backbone of our texting program by providing support to our texting volunteers throughout each campaign.

A big part of why our moderator team was so successful was due to the ownership of the role and #text-questions channel we gave to our volunteer leaders,some of whom went on to be fellows with us. Our moderators would communicate what common questions texters were having or if there appeared to be any issues with the script or ThruText. Volunteers often expressed their gratitude for our moderators and the time they took to answer each question and ensure our volunteers felt supported.

Conclusion

None of what the NextGen Text Team accomplished during the 2020 election would have been possible without our incredible volunteers, volunteer leaders, and fellows. It’s clear to every campaign that there were not enough texts to match volunteer enthusiasm. Our volunteer leaders put in so much effort to lead and moderate training sessions, onboard volunteers efficiently, answer questions quickly when volunteers needed to get information out to voters, and provide constructive and positive feedback to almost every single individual who texted with us.

When our volunteer leaders had questions, though, they always knew where to go. Our smart, creative, and hardworking fellows put in the work each day to make our programs more accessible for both our volunteers and volunteer leaders. Volunteer leaders who struggled with the basics of Google Sheets were sweeping pros by election day. That takes commitment and a mindset that anyone who wants to help can find a place with us, and our fellows brought that each day.

This program was made better by Claire Wiebe, Ellen Parkhurst, Emily Dodge, Liam Davidson, Mari Volante, and Olivia Ghiz. Thank you.