Read time: 9 minutes

2020 starts now, y’all.

Even with a few of us still slightly hazy from election night parties, we’re already digging into what we learned this fall.

Three lightbulbs went off for us this cycle with big implications for 2020—and we want to share:

  • Lightbulb 1: Ignore unlisted and infrequent voters at your peril
  • Lightbulb 2: Digital ads can supercharge a canvass
  • Lightbulb 3: Peer-to-peer SMS supercharges volunteers

Read on for the gritty details, but first, here’s a quick sense of what M+R actually did this cycle.

Short version: we helped America’s leading labor, environmental, and reproductive rights groups break new electoral ground—new roles for volunteers, new digital-field synergy, new peer-to-peer turnout techniques, warm and cold SMS, text parties, digital ad innovations, OMG SO MUCH. And despite some bitter losses and nail-biting recount drama, we were proud to help drive historic wins in a few different ways:

  • Ads: We ran nearly $1.5 million of electoral ads and served over 64 million ad impressions, including personal, proven-to-be-persuasive content from real, everyday people in the communities we touched.
  • Targeting: We pioneered digital targeting to mobilize infrequent voters and the 2 in 5 Black and Latinx people who aren’t even contactable via commercial voter files.
  • SMS: We ran peer-to-peer texting programs for five major progressive organizations representing over 2 million texts sent. (Plus millions and millions of broadcast text messages driving voter registration, electoral awareness, and voter turnout.)
  • Breadth and depth: We ran tightly focused efforts for some clients and everything-but-the-kitchen-sink programs for others across digital acquisition, persuasion, and turnout, SMS, and volunteer mobilization. For example, as part of a major effort to mobilize Black, Brown, and white working-class voters who often get ignored by campaigns, we ran digital ads and SMS for SEIU in Florida and across the Midwest, and digital ads for Planned Parenthood in Florida, Michigan, and Nevada. We helped For Our Future run a test on how best to target infrequent voters of color. We ran two ballot initiatives in Montana—returns are still coming in, but both are on track to win. And plenty more.
  • Moolah: Of course, we also raised millions of dollars for nonprofits that helped support voter education and turnout.
  • Skin in the game: In addition to our client work, we made sure everyone on staff had the time to follow their own passion for our democracy by providing an extra half-day of Election Leave.

Ok, now for those lightbulbs. We’ll obviously know a LOT more as voting data emerges, but we’re already rethinking some things for 2020 based on a few surprising findings.

Lightbulb 1: Ignore unlisted and infrequent voters at your peril.

Academics have found that one in five voting-eligible people are “unreachable”—either unlisted or mislisted on commercial voter files like the VAN and Catalist. That doubles—doubles!—for Black and Latinx voters. But until now, all we had was studies—not hard data from an actual voter contact program.

Now we do. We ran two digital recruitment programs for different organizations, using different content and messaging, both primarily targeting voters of color. In one, 25% of potential voters we found were unlisted or mislisted. In the other, it was 33%.

Traditional targeting doesn’t reach these voters—ad platforms likely match only 45-60% of a given segment of the voter file (fewer as you go for more infrequent voters). And that doesn’t count the additional 25%-33% who aren’t on the file in the first place. That means you’re leaving out a majority of your target audience if you’re only matching to the voter file.

So, instead of relying on traditional electoral targeting strategies, we tested various lookalike and modeled audiences, overlaid with key modeled zip codes, to identify, recruit, and turn out unlisted/mislisted and infrequent voters—reaching people the voter file can’t reach (and that most campaigns don’t even try).

We also found something that disproved countless skeptical consultants: unlisted, mislisted, and infrequent voters are NOT disengaged. Turns out these voters respond to text messages at rates higher than frequent voters: in one program, they made up 52% of recruits but drove 70% of our actions. In another analysis, unreachable and unscored voters drove more SMS responses than listed and scored voters!

We’ll see how the actual returns shake out, but bottom line: Many campaigns are writing off an active and engaged swath of the electorate who are disproportionately Black and Latinx, fueling a cycle of non-contact and—surprise!—non-voting. Institutional racism in the flesh.

You can’t simply show up at election time and ask for their vote. Instead, start with long-lead organizing around issues that matter in their daily lives. Opt them into ongoing SMS streams to organize them, turn them into volunteers, and get them to the polls. After the election, instead of being left with nothing, you’re left with real relationships—ones you can use to organize at scale and build real long-term power.

Lightbulb 2: Digital ads can supercharge a canvass

In a program with For Our Future to recruit and organize infrequent voters of color in key Florida counties, we tested digital ads and canvass versus canvass alone.

Across all counties, we found a 12% bump in canvass RR. In Broward County, arguably one of the crucial political geographies in the nation, the bump was 32%. (These were relative bumps, not percentage point bumps—i.e. in Broward, ads increased our rates from 17% to 25%.)

This is just one test. More are needed. But it suggests that digital ads can make a BIG impact on canvass effectiveness. Since canvasses are heavily resource intensive and costly, an investment in digital ads can ensure that you get a lot more out of that canvass work. If you’re knocking 500,000 doors, bumping a 17% response rate up to 25% means 45,000 more responses—you’d need to knock an extra 265,000 doors to reach that without ads!

Of course, it still might not be worth the money to run an ads program solely to boost a canvass. But those ads will also be, y’know, recruiting and persuading and turning people out.

At the very least, if you’re running a canvass and digital ads, and they’re not tightly coordinated in terms of geography and timing, you’re probably pouring money down the drain.

Lightbulb 3: Peer-to-peer SMS supercharges volunteers

Phonebanking and canvassing are vital to GOTV, but let’s face it: Most normal humans do not actually enjoy calling up strangers or knocking on their doors. (Judging by what my coworkers did with their Election Hours, M+R staff seem to be a notable exception.) No wonder peer-to-peer texting had a breakout year this cycle—MoveOn alone texted 20 million people in the final week.

The volunteers we organized in this cycle really ENJOYED texting. 25-30% of them signed up for multiple “shifts.” They had meaningful conversations with voters thousands of miles away. They formed bonds with each other in Slack and closed Facebook groups. They recruited and shepherded others through the process, cutting down on the need for staff resources.

With that positive experience in mind, we’re working to take advantage of these fired-up election volunteers to drive our clients’ digital organizing successes in 2019, starting NOW. We can’t let these folks go cold—so we’re shifting these volunteers from texting the voter file into texting other volunteers to promote higher-bar action. These are the seeds of a true distributed organizing program for some organizations.

We also learned plenty that will inform these programs in 2020:

  • Most people drop off in the transition from signup to their first actual text—we found flake rates of 50+% in some cases. That’s where small interventions can have the biggest impact. We’ve optimized this transition for 2020 by streamlining training and support, doing extra follow-ups with people who haven’t yet signed up for a shift, assigning lists to texters well before their shift starts, and doing it all with volunteer leaders to reduce staff time and ensure a more personal touch. (If you can get them signed up for a shift and assigned names, you’re golden: volunteers text 90% of the voters they’re assigned.)
  • To text 100,000 voters, we need 320 people to sign up. This conservatively assumes ~20% of your vols do an additional shift, that you’ll see a 50% flake rate between signup and first text, and that each of those 160 texters texts 500 people. Organizer math FTW! This could, of course, change in future cycles as more people become accustomed to the tool (and as we optimize these programs to the hilt).
  • We should embrace the fact that our programs saw big bumps in the number of texts sent in the final few days—partly because our efficiency increased and we had all those repeat-texters! Of course, that’s when text is most useful from a turnout perspective, so we’ll lean into that. We’ll work out the kinks with a small group of super-users, then expand RAPIDLY in the week before the election— testing what works with a small volume then doing a ton of last-minute recruitment to go huge at the very end.
  • We should always give volunteer leaders specialized roles (i.e. one person is in charge of helping people get logins, one person is in charge of helping people get contacts, etc). This one is pretty much a no-brainer, but it helped our efficiency a lot, and it’s a good reminder for any distributed organizing program.

Bonus lightbulb! Relational voter turnout

Many of us are frustrated by the transactional nature of electoral organizing—especially in the typical last-minute rush for votes in Black and brown communities. And, as much as we love peer-to-peer texting, there’s no getting around the fact that it’s coming from a stranger.

Relational voter turnout (RVT)—volunteers contacting their own personal networks for turnout—just might be the antidote. We worked with 2 national organizations to roll out this proven tactic across the country, helping volunteers invest in their personal relationships and deepen their civic engagement for the long haul.

We’ll report more on RVT soon, but I can’t resist sharing a few quick takeaways. While people can participate in RVT programs at home in a distributed way, we saw higher rates and better retention when volunteers get together and do outreach in person. We also found some major constraints working with most existing RVT tools, especially with communities of color, since they require volunteers to bump their personal contact lists against the voter file. This adds time and complexity to an already intensive task. And that’s to say nothing of volunteers’ friends who are unlisted and mislisted. It’s likely that as political polarization increases and people’s personal networks become more politically homogeneous, these tools may not work as well for communities of color as simply asking them to contact their friends and tracking those contacts.

OK, that’s the scoop for now! Time to go follow the recounts.