Doing data for good
Updated: Jan 28
I struggle to talk about my job. If I'm talking to someone who works in progressive politics and I mention that I'm the Data Director at Sunrise Movement, they immediately get the picture. Boom. No further conversation needed. But when introducing myself to literally anyone else I spare myself the struggle and just say that I work in politics. When I do decide to go down that path, I struggle to explain why I do not have the same skill set as my tech counterparts, why the hardest part of my job is not coding, and why it really is okay that I do not have admin privileges in my own data warehouse. Ultimately, it is not a very productive conversation.
I think I can do better than all my failed attempts to explain my job on the fly. I love what I do and have a real sense of pride about my work. And I want others to see this work the way I see it: a deeply challenging yet rewarding field that is separate from the for-profit data world.
What is Progressive Data
The field I work in is called many things, but I call it progressive data. I work in an ecosystem of campaigns, consultancies, advocacy orgs, pollsters, research institutes, and tech companies that—more or less—work together to advance to progress. Each of these organizations values a different set of skills and expertise for its data staff. I have only ever worked at advocacy orgs, and will be speaking from my personal experience. Far smarter, more experienced data professionals have written about this very topic and you should absolutely read their words.
The for-profit world has shaped our understanding of what it means to be a data professional. A survey of data job postings yield titles like Machine Learning Engineer, Data Scientist I, Data Analyst, Junior Data Engineer. These are (mostly) not the titles you would see at an advocacy org like mine. There is a whole smorgasbord of bootcamps that promise to take aspiring data scientists, push them through intensive coursework, and spit them out into 6-figure roles at top companies. I attended one of these bootcamps when I was first exploring my interest in data. I remember pleading with the career services expert that had been assigned to me that I wanted to leverage my data skill set for good, but that I didn't know what kinds of jobs to look for. They didn't either. And now that I do have that "data for good" job, I hardly use any of the skills I learned about in my Data Science™ bootcamp.
Progressive data is its own field. We have our own set of desired skills and experiences, tech stacks, and jobs. A resume that would guarantee someone a senior role at a tech company would not necessarily have the same luck in our world. So what is it that makes progressive data so different from it's for-profit counterpart?
Organizing is key
There's a saying that it's easier to teach an organizer how to code than it is to teach a programmer how to organize. Organizing domain knowledge is often more valuable than coding experience. I have managed several people who had little to zero coding experience before joining the data team, and all of them were successful in their roles. Why do we value organizing experience so much? First, just as domain knowledge is a key pillar of data science, progressive data professionals must understand real-world problems they are attempting to solve. You will have a more difficult time building a report for your text banking team if you do not understand how text banking works.
But there's a second, more subtle, reason that organizing experience is so crucial in our field. If you have ever spent time knocking doors, making phone calls, or running a training program, you understand that tech is not the limiting factor to success for progressive causes. There is only so much that optimized data pipelines, machine learning algorithms, and world class data analytics can do to really build power on the left. We are not going to hack our way to a better world. An app is not going to save us. We build power through organizing - by having one conversation, and then another. No amount of tech will ever come close to having the impact as doing the damn work.
And when you have this mindset that organizing comes first, your role as a data person in the progressive space shifts. The focus of my work is not about building a data product. Data teams that are product teams, as I often think is the case in the for-profit world, have the luxury of getting to take their time to build robust, code-reviewed, well-tested infrastructure. The work of running a data team is divided among many analysts, engineers, data scientists, and more depending on the size of the company.
I do not have that luxury. At it's core my data team is a service team: I provide insight to organizers about the efficacy of their work. Often the infrastructure I build in our data warehouse only needs to exist for a few days to a few months. In the case of campaigns, your code only needs to hold together through election day—it can all come crashing down the day after and it does not matter. Our work is fast-paced and constantly changing. I could build out the reporting for our trainings program one week and then the next day learn that the trainings team has pivoted to a different program and they need an entirely different solution. That means at the end of the day if I have to hard code data into a script or copy and paste values manually into a dashboard in order to meet the needs and deadlines of my organizers, I'm going to take those short cuts.
When I speak to people who work in tech, I struggle with this concept the most. Being a good programmer will only take me so far in my job. I could spend weeks optimizing my code base for efficiency, building out unit tests for core parts of my pipelines, and migrating off Redshift to something less frustration-inducing. All these best practices would certainly make me and my data team very happy, but would barely translate into real world outcomes. At the end of the day I just do not need my team to be at the cutting edge of technology.
What, then, makes a skilled data professional on the left? Extensive knowledge of the voter file and its hundreds of models; VAN and Google Sheets wizardry; understanding of campaign finance law and its impact on data compliance; expertise with platforms such as Spoke, EveryAction, ThruTalk, Hustle, Mobilize etc.; and prior experience as an organizer to name a few.
This desire for progressive-specific data experience is why so many data professionals on the left get their start as organizers. A natural career path for an organizer is to first get thrown on to a campaign where they become VAN experts. Or maybe they become the Google Sheets guru of their campaign and become the go-to tech person because there is no dedicated data person on the campaign. Then they might get their first data-specific job as a VAN administrator and learn to code on the job. At this point this hypothetical (yet all too real) person is qualified for all sorts of data jobs on the left.
Okay, so now will you tell me what you do?
As the Data Director of the United State's largest youth climate organization, I answer the question: are we winning? On a day to day basis this looks like working side by side with organizers to set ambitious, measurable goals for new programs; leading a team of data managers to develop data collection systems, training organizing staff in those systems, and then enforcing the use of the systems; writing a whole lot of SQL to transform raw source data from tools like EveryAction, Spoke, ThruTalk, and Mobilize into something that has meaning to our organizers; and then reporting back the insight we gained from collecting data through dashboards, emoji-filled Slack posts, and... okay it is mostly a lot of dashboards.
When new programs pop up, it is my job to build the infrastructure to support that program. A lot of the time it looks like ensuring the program is using EveryAction Online Forms correctly and collecting all the standard data we collect in my organization: race, gender, birth date (not age!), and socio-economic class. In the case of Sunrise School, I worked with organizers to "hack" EveryAction to suit their needs. We used EveryAction's tags feature to store information about facilitator group assignment and implemented a naming convention that would tie multiple separate events together into one "course." If we have a new PAC related program, it is my job to ensure we have bought and moved data appropriately in our CRM. We work with our lawyers to ensure our contact data is priced accurately and that we are following campaign finance law in the transfer. None of these activities are technically "data'' related—no coding is involved. But thinking through the systems needed for programs to work is an essentially part of being a Data Director.
When I do code, 95% of my work is done in SQL. I consider myself primarily an Analytics Engineer, meaning along the ELT pipeline I'm mostly concerned with the T part. I write SQL to transform raw source data to give it meaning about our programmatic work. When I was primarily doing field data for our electoral work, this looked like writing the code to support a dashboard that informed my organizers whether or not they were hitting their goals for their phone and text bank programs. In addition to needing to know the number of texts and calls we make, we also delve further into the drop off rate from RSVPing to a phone/text bank and attending, calls made per people on the dialer, responses to the survey questions in the text and phone scripts, and so much more. Every one of those questions can be answered by a SQL script, and it's my job to write it fast and put the results somewhere where they will be seen. But that is just one program, and I am often responsible for doing this work for many, many programs at once.
While a lot of my work is reactionary and in service of the day to day demands of my staff, there are times where I get to be visionary in my work. I am currently building out infrastructure that will give insight into the scale and quality of Sunrise's movement. The challenge is to take all the data an individual might generate during their activity with our movement—event attendance, hub affiliation, role network position, donation history, trainings completions—ascribe these data some kind of meaning, and then throw it all together in a way that tells us something about the strength of our movement. Through this work I am trying to answer questions such as: what first entry points into the movement lead to more engaged or retained members? Does engagement differ across race, gender, class, or age? What "path" leads to the most retained members? This project is both technically challenging and pushing the limits of my ability to create meaning from abstraction.
Is progressive data really that different from tech?
I have spent the bulk of the article outlining the difference between the for-profit and nonprofit tech worlds. If I sound defensive, it is because I, and so many of my peers, receive an onslaught of emails from people who work in tech asking to volunteer on or consult for our data teams. Most of these people have never knocked a door in their life. They usually do not understand that their offer would potentially displace a junior hire that we can actually invest in, a person who will stay in progressive politics and build out our talent pool. They approach us with the misbelief that we as people who work at nonprofits must be "scrappy" and "disorganized" and the hubris that we need the help in the first place.
These are good people. And I want to help these people find meaningful work, volunteer or otherwise, because there is no shortage of work to be done in the progressive movement. I think we do ourselves a disservice when we restrict access to this field. The scope and purpose of our movement necessitate the need for top tier talent. I myself work in climate, and the urgency and weight of the climate crisis demands the effort of a full mobilization. We need all hands on deck if we are going to pass a Green New Deal and create millions of good jobs. And that is just one issue. The progressive movement works on a myriad of issues, from abortion access to criminal justice reform to workers rights. There is a real need for talented data professionals to join us in these fights.
There are some real lessons to be learned from our for-profit data counterparts. I know I personally have become a stronger engineer through the generous mentorship I have received from who works in tech proper. That is not to say that the progressive-specific trainings and bootcamps are not effective in training up our field—everyone I manage has attended one of these training sessions at my recommendation. It is only my belief that there are lessons to be learned from people whose job it is is to write solid, robust code.
At the same time, there are very real reasons why we do not advertise our jobs outside of our insular jobs boards and keep our list servs restricted. Apart from the whole Republican mole thing (ew, moles, go away), everything I wrote above bolsters the argument that our field is independent from tech with a set of desired experiences that come from holding some other role in the progressive movement beforehand.
I do not have a solution for this. Only a hope that we can continue to wrestle with these ideas and that if you want to make the move to progressive data, that you join us in making a few phone calls first.
This article was made possible by the generous content-editing skills of Maya Handa