There is a small mystery at the heart of nonprofit marketing. The people with the least patience for waste often work in a business that produces an astonishing quantity of it: ignored emails, skipped videos, unopened envelopes, abandoned donation forms. M+R, a Washington-born communications agency founded by organizers in 1991, has spent three decades treating this apparent waste as evidence. A failed channel is not merely a disappointment. It is a clue.
That distinction explains what M+R actually does better than its long menu of services. Yes, it makes ads, raises money, pitches reporters, builds brands, designs websites, sends texts, buys media and helps coalitions move lawmakers. Its clients are nonprofits and public-interest groups - Feeding America, the International Rescue Committee, the National Education Association, NPR, the Sierra Club and the MacArthur Foundation among them. But the interesting product is the learning system around all that work.
The least polished ad in the room
In 2025, four unions - AFSCME, AFT, NEA and SEIU - needed to make federal cuts to Medicaid and public education feel immediate in 18 Republican-held districts. The usual political machinery was already grinding. Voters had seen the attacks. Another grave narrator over ominous music risked becoming expensive wallpaper.
M+R and its partners spent $2 million on Put Families First, a storytelling and mobilization campaign built around 39 people describing what the cuts would mean in their own lives. Production was deliberately plain. The campaign then used dynamic creative to turn those stories into more than 2,000 localized ad variants, aimed by state, district and lawmaker.
The first thing to fail was not distribution. It was the assumption that political advertising must look like political advertising. In head-to-head testing, the candid stories beat a traditional attack ad. M+R changed the creative model accordingly: fewer scripts, less lacquer, more room for the speaker’s own cadence. The campaign recorded 267,000 shares and other paid-ad engagements.
Polish is useful until it becomes evidence that somebody else is talking.The practical lesson of Put Families First
Every exhausted channel is trying to tell you something
The same pattern appears in M+R’s fundraising work. The International Rescue Committee’s donor acquisition on Meta began showing diminishing returns. Tweaking the existing feed was not enough. M+R spread the program into audio, podcasts, online video and connected TV, while keeping a common measurement framework. In the following year, those channels became top performers. During major emergencies, the agency says it more than doubled the annual media budget while holding or improving return on ad spend and cost per donor.
A stalled channel changed the media mix. A lapsed donor file changed the message. After Hurricane Ida, Second Harvest Food Bank of New Orleans and Acadiana had a large group of emergency donors who had gone quiet. Instead of opening with another hard ask, M+R sent a cultivation message and survey about what had first moved them to give. The treatment lifted response rate 34 percent and average gift 14 percent.
The cleanest example may be the Museum of Science in Boston. Its annual Pi Day fundraising had become more about pi and a plush mascot than the museum’s mission. M+R tested images, taglines and mission statements with ticket buyers, email subscribers, a lookalike audience and board members. Then it brought the museum’s departments into one campaign plan. Five email appeals raised nearly two-and-a-half times what ten had raised the year before - a 390 percent increase per appeal. The copyable lesson is almost rude in its simplicity: test what the audience thinks the story is about before making more of it.
The ruler is part of the business
Many agencies publish reports. M+R’s annual Benchmarks study is more entangled with the work than the phrase “thought leadership” suggests. The 2025 edition combined data from 216 nonprofits into 115 charts covering email, advertising, web traffic, mobile, social media and influencers. In 2026, the study reported that average online revenue rose 15 percent in 2025, while one-time giving grew faster than monthly giving - a break with the familiar pattern.
This gives M+R a public ruler for the category and a private question generator for its own product team. When the reporting tools it wanted did not exist, it built Scout Reports and Scout Ads. Scout Quest followed as a data cooperative. A participating nonprofit submits its file and learns which people already on that file are clicking, acting or donating elsewhere. It does not receive somebody else’s names. The distinction matters: this is enrichment, not a list swap.
The model is neat because free research helps a nonprofit diagnose a problem; agency work helps solve it; and paid tools make the answer repeatable. In 2026, M+R added Scout AI, using it to examine five years of email content from 12 nonprofits. This is not a software company disguised as an agency. It is an agency that keeps discovering small pieces of software inside recurring client problems.
A progressive shop has to show its work twice
M+R says it only works with clients aligned with its values. That narrows the market on purpose. It also raises the standard internally: a firm advising unions and justice organizations will be examined by its own staff with unusual care. M+R’s union is therefore more than a culture-page ornament. In November 2025, members unanimously ratified a contract with 5 percent annual raises, a 36-hour workweek, expanded leave and an agreement that generative AI would be a tool rather than a replacement for union staff.
The company is remote-first, with offices in Washington, New York and Oakland and staff living across 29 states. M+R reports 6.5 years of average tenure; 54 percent of its leadership has been there more than a decade, and eight employees are “boomerangs” who left and returned. Institutional memory is not merely pleasant in a test-and-learn business. It is part of the database.
What to steal, carefully
The transferable M+R playbook has five moves. Begin with the behavior that is failing, not the channel you happen to own. Ask the audience a question before writing the next appeal. Put a control group next to the clever idea. Measure the action that matters, whether that is a donation, a lawmaker email or a retained donor. Finally, publish enough of the result that the next team does not begin from zero.
You have enough audience volume to run a real test, a meaningful conversion to measure, and permission to change the message or media mix when the evidence disagrees.
The sample is tiny, the crisis leaves no test window, attribution is fragmented, or an organization wants the aesthetics of experimentation without accepting an unwelcome result.
There is one more constraint. M+R’s advantage compounds because it sees patterns across many sizable nonprofits. A small local organization cannot reproduce that field of vision by buying a dashboard. Benchmarks themselves warn against crude year-to-year comparisons because the participant pool and methodology change. A benchmark is a compass, not a performance review.
That is why the most useful thing to copy is not M+R’s technology, its media budget or even its fondness for plus signs. It is the habit of treating response as a conversation. The unopened email is saying something. So is the low-budget video that outruns the expensive spot. Somewhere inside the clicks, gifts and silence, the audience has already written the next brief.