Search
  • Brittany Bennett

A Practitioner's Guide for Measuring Movements


I study a movement for a living. In 2020 I was hired at the Sunrise Movement to bring a data-driven lens to our organizing, and since then I have had the privilege of building out their data infrastructure from scratch. Along the way, I learned a thing or two about what metrics matter (and do not matter) when you are working to build long term people power.


I want to offer a first person account of how one person measures power at their organization. I do not have access to fancy research methods, and instead have crafted a set of metrics and analyses that can be done by anyone with the right tools and skillset. A kind of everyman's toolkit for getting started with power analysis. This guide encompasses as set of analyses that you can do right now, with your data, to uncover insight on the success and shortcoming of building people power in your organization.


However, I must admit that I am far from the only person doing this work, let alone an expert on this subject. Researchers like Harhie Han, Elizabeth McKenna, and Michelle Oyakawa (and more) have been studying power in movements far longer than I've been working in progressive politics. If you're looking for a more in depth, well researched account of how to measure power in organizations with a sample size larger than one, I highly recommend Prisms of People and the work of P3 Research Lab.


The following is an offering. A template for measuring power in a movement with 5 metrics. These, of course, do not encompass all the analyses you could do to measure power. Far from it, they are, instead, five very potent metrics for assessing people power that anyone with a data warehouse and basic SQL skills could replicate for their own organization or group.


Finally, we do not collect and analyze data for the fun of it. There should be no input without output. If we are going through the trouble to collect all these data in the first place, we must leverage it to inform our organizing. I tie every metric below to a set of real world decisions that you could pose to your organizers. These numbers should not just sit idly in a dashboard, but instead be used to drive your movement.


Below is my practitioner's guide to measuring power in movements.


All the charts and tables below use fake data.



The One Table to Rule Them All


Before I jump into the metrics, I need to talk about the summary table that feeds it all. The root of this analysis stems from what I call a "full action history" of our base. This is table of every person in your movement and all the actions they have taken, ranked in the order the timestamp they took their actions.


The beauty is that you get to define what an action means for your organization. For me, an action could be an email sign up, a donation, signing a petition, calling you representative, attending a national event, joining a training, and more. In my example below (Table 1) you will see that I have included a tag for each action, which buckets actions into broad categories (like Petition or Ad). I further summarize the action into Lowbar (or online) or Highbar (event RSVPs) actions.


Table 1. A sample of what a full action history might look like for your organization.


Getting your data into this shape is easier said than done. For my case, just scoping out all possible actions was tricky enough, let alone normalizing each source into something that could be unioned together with the other action histories. I also encountered a duplication behavior—responding to an ad and other digital actions also signed the person up for our email list with the exact same timestamp, which interfered with my ability to assess a person's true first action.


But if you can wrangle your data into a table like the above, you open up the door for a wide array of possibilities for analyzing power in your movement. This table can be cut and sliced a myriad of ways into any number of metrics.



Five Metrics for Assessing People Power


1. Total engagement of the base month to month.


How active is our base at any given point in time? What is the quality of that engagement? How does that change over time? With the table created above, we can use a simple "group by" to tally up the total actions taken for any timeframe we may want: week, month, quarter, year. Below is an example of what this might look like (Figure 1). We can quickly see in this hypothetical example that we had just over 10,000 actions taken in April 2020, compared to near 15,000 actions taken in April 2021 (suggesting that maybe our movement is growing?). We also observe an expected increase of actions taken near an important election, and a lull during the winter months that follow.



Figure 1. An example of how to measure engagement in a movement by tallying the total actions taken per month.


Given that we "tagged" our actions with "Lowbar/Highbar", we could also add a quality to the chart above. How many Lowbar versus Highbar actions are taken in any given month? Does that composition change ever, and why? If you want to get extra fancy, you can weight the data by assigning a set of points to each kind of action. Some CRMs, like EveryAction, has this feature built in, but I find it much more rewarding to have full control over the logic and weighting scale on the backend. And, if you have data that does not sync to your CRM, you can still include it in your analysis.


Decisions made from these analyses:

  • Do we need to change our programming or asks to be more relevant or interesting to our base?

  • Are we providing our base with enough ways to take action? Should we increase the number of asks this month?

  • What barriers might exist to taking action? How can we make volunteering with us more accessible?

  • Should we repeat any popular actions? Not run any actions that didn't perform well?


2. New members to the movement

With the "rank" column in the table above, we can track a person's first action with our movement. So here we can look at metrics like the quality of that first action (with either the engagement points discussed above or the Lowbar/Highbar summary), what actions brought in the most new people in any given time frame, and the popularity of first action "types" (donation, petition, ad, email sign up, mass call, hub meeting, training, etc).


Figure 2. An example of how to measure the strengths and weaknesses of how your movement is bringing in new people.


Above is an example of how you can cut the original table to analyze how your movement is growing. We see that most people's first action this month is signing up for our email list, followed by donations, but then we see that our latest Orientation Training brought in over 400 people! Maybe we should take a deeper look at that Orientation Training to see if our organizers have anything to say about its success...


Decisions made from these analyses:

  • Should we repeat any actions to bring in more people?

  • Do we need to tweak the programming of any actions to resonate more with the people we're attempting to reach?


3. Lowbar to Highbar conversion

Now we can start to take a look at absorption. Let me be the first to say that measuring absorption is difficult. If you can pin point what exactly you mean by "absorption," well...you still have to find a way to model it in your code base.


In my case, I have three main ways of assessing absorption, which take us to the end of this list. The first metric is looking specifically at Lowbar to Highbar conversion. That is, of the people whose first action was a Lowbar action, how many of them end up taking a Highbar action? And of that subset, how many of them do so within 30 days?


For me, while the 30 day window is somewhat arbitrary, time bounding this metric is key. In our case, we are interested in activating our members within a certain time frame of their first action.


The good news is that with our "full action history" measuring time between actions is fairly straightforward!


  1. Make a table from the "full action history" where rank = 1 and action_type = Lowbar

  2. Make a table from "full action history" where rank > 1 and action_type = 'Highbar"

  3. Make a table to grab the earliest Highbar action from Step 2

  4. Join Tables 1 and 3 together and compare the dates of the Lowbar action with the earliest Highbar action


At the end of the day, you should have a baseline Lowbar to Highbar Conversion Rate that is the average for your whole data set (or the average for the last year) and the Lowbar to Highbar Conversion Rate per month (grouped by the date of the first action) for both the unbounded and time bounded metrics.


Decisions made from these analyses:

  • What interventions should or can we make to encourage more people to take Highbar actions?

  • What barriers and limitations can we remove from our culture, programming, or structure?

  • What new programs should we create that will resonate with our base?


4. Repeat actions

We can also look at absorption by the spread and frequency of people taking repeat actions with our movement. First, we can look at the distribution of the number of actions taken per person (Figure 3). If you make a plot at this at home, it is very normal if the vast majority of your base as only ever taken one action.


Figure 3. An example of a distribution of the number of actions taken in a movement.


Once upon a time I used the percentage of people who took at least a second action with us as the definition for absorption. And while I still think this is a useful metric, I had a meeting with someone who told me that the preferred metric here was to look at whether someone took a second Highbar action. Since our "full action history" is set up the way it is, tracking both of these metrics is relatively simple! For my use case, absorption occurs at the third Highbar action.


Decisions made from these analyses:

  • Should we run a reengagement campaign, and on who, to encourage more repeat actions

  • How large is our crowd? What kind of engagement on Lowbar actions can we expect, and how does that impact our strategy?

  • What are common actions among people who take multiple actions, and can we do more of these?


5. Member Absorption


Finally, we can analyze absorption in our movement be calculating the rate at which people end up becoming members. As with the Lowbar to Highbar conversion metric, we also want to timebound this metric. For my use case, we use the arbitrary timeframe of 90 days as our ideal window for becoming a member from a person's first action.


You may be in a similar situation as I am where there actually is not a clear definition of a member for your organization. If you do have a clean table you can pull from of all your members, I am very jealous. Instead, we made up our own definition of a member: someone who has taken at least 3 Highbar actions.


Decisions made from these analyses:

  • Are our programs effective at absorbing people as members, why or why not? And what can we do to improve them?

  • Which of our programs are the best at absorbing members long-term, and should we repeat them? and how?

Demographics


Okay I could not just end things here with only 5 metrics. Like many of you reading this, I am committed to building building multi racial, cross class people power. Not only do I measure everything I listed out above, but we intersect each metric with a racial and class analysis.


So, for example, we look at not just how many new people are entering our movement, but how many of those people are BIPOC and/or working class. Of the people that became members, is there a difference across race? across class? We measure the strength of our programs not only by the scale of their absorption and retention, but how these programs are building multi racial, cross class people power.


If you are like me, you also have a difficult time doing these analyses because of the lack of demographic data available early on in your ladder of engagement. You may be able to make due with a match to voter file. Since I work at a youth organization, we instead rely on self reported demographics to feed our race and class analysis. Knowing the true demographic make up of our larger crowd and member is one of the greatest challenges of this work.



Conclusion

If you have gotten to the end of the article and are teeming with "But wait! What about X?," do not worry -- so am I! There is so much I did not cover (date of last action for reengagement campaigns, ladder of engagements, geographic distribution of members, assessing local chapters/hubs strength), but also so much I am still learning. No doubt 6 months from now I will be itching to make a Part 2 to this blog. But for now, I hope this leaves you with enough to get started with measuring people power at your own organization. And if you want to chat more about this kind of analyses (and more) feel free to reach out to me https://www.brittanybennett.com/contact


This blog was written to accompany my 2021 Research + Experimentation 9 Convening talk: How to Measure a Movement.

908 views0 comments

Recent Posts

See All