🏛️ Issues and Advocacy Training
Courses › Advocacy Social Media Strategy: An Evidence-Based Playbook for Followers and Email Lists

Lesson 10 of 10 · about 46 minutes

Operations and the Plan: Running a Page Network for Twelve Months

You'll be able to

  • Build an editorial calendar whose cadence per platform is set from benchmark posting data rather than habit
  • Assign roles and platform permissions across several pages using Meta's Business Manager roles and Page tasks, and keep a role matrix current
  • Run an approval workflow that checks documents, altered media labels, private information and paid partner disclosure before anything posts
  • Write a crisis protocol that reflects what the stealing thunder and situational crisis communication research actually supports
  • Secure every account with phishing-resistant or app-based two-factor authentication and admin hygiene, knowing the blocking rates from Google's 2019 research
  • Set a budget at volunteer, small nonprofit and funded campaign scale, and justify a 12-month plan with milestones a board can check

Start here: the short version

  • A page network runs on four documents: an editorial calendar, a role matrix, an approval checklist and a crisis protocol. Write them before the campaign, not during it.
  • Set cadence from data. Nonprofits in Rival IQ's 2025 report posted a median 5.5 times a week on Facebook and 4.2 on Instagram, and M+R's 2026 nonprofits sent 50 emails per subscriber a year. Start below the medians and earn your way up.
  • Security is cheap and failure is total. Google's research found on-device prompts blocked 100 percent of bots, 99 percent of bulk phishing and 90 percent of targeted attacks. Every admin on every account, no exceptions.
  • When you have bad news about yourselves, disclose it first. Small experiments (168 participants) found ex-ante disclosure protected reputation better than waiting. The "respond within an hour" rule has no experimental support that we could find.
  • The worked plan below takes a fictional three-page Idaho network from 400 to 2,500 email subscribers in 12 months on a $7,200 budget, with every rate drawn from a cited benchmark and every milestone on the dashboard from lesson 08.

Strategy fails in operations. The theory of change from lesson 02, the platform choices from lesson 03, the content evidence from lesson 05, none of it survives a volunteer who posts an unlabeled AI image at 11 p.m., a treasurer who loses the only admin login, or a coordinator who quits in month seven because nobody set a cadence a human could keep. This lesson is the operating manual, and it ends with the plan that puts the whole course to work.

The editorial calendar

The beginner's version of a calendar is in Content Calendar in the social posts course. For a network of pages, the calendar does three more things: it assigns every post to an objective from the metrics table in lesson 08, it schedules cross-posting so the same item does not land on three pages in the same hour, and it reserves capacity for the unplanned.

Cadence should come from benchmarks, adjusted downward for what your team can sustain. Rival IQ's 2025 report, built on 150 randomly selected companies per industry with 25,000 to one million Facebook fans, found nonprofits posting a median 5.51 times a week on Facebook, 4.18 on Instagram, 1.42 videos a week on TikTok and 5.68 posts a week on X. Those are mid-sized organizations with staff. On email, M+R's 2026 benchmarks for 180 nonprofits put the median at 50 messages per subscriber per year: 31 fundraising appeals, 9 newsletters, 2 advocacy alerts and the rest engagement or other. A small advocacy network is not a national charity with a fundraising calendar, and the plan below sends far fewer appeals and more advocacy and newsletter mail. The point of the benchmark is to know where the middle is before you choose to be below it.

ChannelBenchmark median (source)Volunteer network cadenceSmall nonprofit cadenceFunded campaign cadence
Facebook, main page5.51 posts a week, nonprofits (Rival IQ 2025)3 a week5 a week7 a week plus Stories
Facebook, issue and chapter pagesSame2 a week each3 a week each5 a week each
Instagram4.18 a week, nonprofits (Rival IQ 2025)2 a week3 a week plus Stories5 a week plus Stories
YouTubeNo posting benchmark verified1 long video a month2 a month plus 1 Short a week1 a week plus 3 Shorts
TikTok1.42 videos a week, nonprofits (Rival IQ 2025)Skip or 1 a week as a test2 a week4 a week
Email newsletter9 newsletters a year median (M+R 2026)Every 2 weeks (26)Weekly (52)Weekly plus alerts
Email advocacy alerts2 a year median (M+R 2026)As hearings require, 6 to 1010 to 15As needed, with segments
Email fundraising31 a year median (M+R 2026)2 to 46 to 812 to 20

How to: build the network calendar

  1. Open a shared spreadsheet with one row per planned post and columns for date, page, platform, format, objective (from lesson 08's table), owner, status (draft, approved, scheduled, posted), link with UTM tag, and the primary metric to be filled in a week later.
  2. Enter the fixed dates first: hearings, filing deadlines, election day, the 60-day and 30-day electioneering windows under Idaho Code 67-6602(6), and your own events. Everything else is scheduled around them.
  3. Set each page's weekly slots from the cadence table and fill them two weeks ahead. Leave one slot a week per page empty for news.
  4. Schedule cross-posts with a gap of at least a day between pages, and give the issue page the first run on its own issue. The architecture rules in lesson 04 explain why the hub should not always lead.
  5. Batch production: one afternoon a week for graphics and video, one for writing and scheduling. Kohavi's advice about running tests in whole weeks applies to production too; a steady weekly rhythm is easier to sustain and easier to measure.
  6. Hold a 20-minute weekly review with the dashboard open. Fill the metric column for last week's posts, move anything unposted, and record what was dropped and why.

Roles across pages

A network with three Facebook Pages, two Instagram accounts, a YouTube channel, an email tool and a website has at least eight places where somebody holds the keys. Meta's structure for this is Business Manager. Its documentation describes two business roles: Admins control everything, including modifying or deleting the business and adding or removing people, with read and write access to all connected assets; Employees can view business settings and have read-only access to assets until an admin assigns them tasks on specific Pages or ad accounts. Finance permissions are separate. On each Page, the Pages API overview lists the tasks: MANAGE (assign tasks to others), CREATE_CONTENT (publish as the Page), MODERATE (respond to and delete comments, and handle a connected Instagram account), MESSAGING, ADVERTISE (create ads and unpublished posts), ANALYZE (view Insights and see which admin posted what), plus MANAGE_LEADS and VIEW_MONETIZATION_INSIGHTS. A person given Admin access to a Page in the interface can do all of them.

The principle is least privilege, which security people have preached for decades and which also solves an organizational problem: a volunteer who can only CREATE_CONTENT on the chapter page cannot accidentally post the chapter's draft on the main page, run an ad from the wrong account, or remove the coordinator. The matrix below is a template.

RoleBusiness ManagerMain pageIssue pageChapter pageAd accountEmail toolWebsite
Board chair (backup keyholder)AdminAdminAdminAdminAdminOwnerOwner
Communications coordinatorAdminAdminAdminMANAGE, CREATE_CONTENT, MODERATE, ANALYZEAdminAdminEditor
Issue campaign leadEmployeeCREATE_CONTENT (scheduled only)CREATE_CONTENT, MODERATE, ANALYZENoneNoneAuthor on the issue segmentAuthor
Chapter volunteerEmployeeNoneNoneCREATE_CONTENT, MODERATENoneNoneNone
TreasurerEmployee with financeANALYZEANALYZEANALYZEFinanceBillingNone
Comment moderator (evenings)EmployeeMODERATEMODERATEMODERATENoneNoneNone

Two Admins on everything, never one. When the single admin's phone dies or the single admin walks away angry, the page is gone, and recovery through a platform's support process can take weeks if it works at all. The second admin is the board chair or another officer who does not post and whose only job is to exist. Review the matrix every quarter and the day anyone leaves.

The approval workflow

An approval step is where the law and ethics from lesson 09 become habit. It should be fast enough that people use it; a checklist that takes three minutes gets used, a review committee does not.

How to: the three-minute pre-publish check

  1. Document linked? Every factual claim about a public body, budget, vote or person points to a primary source, in the post or the first comment.
  2. Altered or generated media labeled? Any AI-generated or edited image, clip or voice carries a visible label. If it depicts a candidate's speech or conduct and the date is within 30 days of a primary or 60 of a general election, the exact disclosure from Idaho Code 67-6628A is on it, or it does not go out.
  3. Private information absent? No home addresses, personal numbers, family details or private images. Public conduct of public roles only.
  4. Paid or partnered? If anyone was paid or given anything to post it, the disclosure is in the post itself. If it expressly advocates for or against a candidate or measure and money was spent, the "paid for by" identification under Idaho Code 67-6614A is on it and the treasurer has logged the expense for the reports covered in the campaign finance course.
  5. Right page, right objective, tagged link? The post is on the page the calendar says, assigned to one objective, with a UTM-tagged link if it has a link.
  6. Second pair of eyes for anything naming a private person, anything about an opponent's conduct, and anything in the electioneering window. One other admin reads it before it schedules.

Build the checklist into the calendar's status column: a post cannot move from "draft" to "approved" until the six boxes are ticked, with the initials of whoever ticked them. The record of who approved what is also your evidence of good faith if a post is ever challenged.

The brand and voice guide

A voice guide for a network answers one question: could a reader tell which page a post came from, and would they recognize all three as part of the same effort? The guide itself is a two-page document: the names and handles of every page as they must be written; the one-sentence purpose of each; the shared visual kit (two colors, one typeface, a logo lockup per page) with the files in a shared folder; the tone in five adjectives and five anti-adjectives; the reading level; the rules on labeling altered media and crediting images; and a dozen example posts, good and bad. The social posts course's Anatomy of a Post covers the sentence-level craft.

On reading level there is evidence. Unbounce's 2024 Conversion Benchmark Report, from 41,000 landing pages and 464 million visitors, found pages written at a 5th to 7th grade level converted at 11.1 percent against 5.3 percent for pages written at a professional reading level, and reported negative correlations between conversion and difficult words (-24.3 percent), reading time (-19.4 percent) and word count (-18.6 percent). That is landing page copy for commercial offers, not advocacy posts, and it is correlational. It still points where every plain-language guide points: short words, short sentences, one idea per post. Write the water rate explainer so a tired parent reads it on a phone in a parking lot.

The crisis protocol

Crises for an advocacy page come in two kinds. Someone else's: a false rumor, a coordinated attack, a hostile story. Yours: a volunteer posted something indefensible, a document you shared turns out to be wrong, a board member has a conflict you did not disclose. The response rules differ, and the evidence is thinner than the consulting industry pretends.

Coombs's Situational Crisis Communication Theory, set out in Corporate Reputation Review in 2007, sorts crises by how much responsibility the audience attributes to the organization: a victim cluster (rumors, attacks, disasters), an accidental cluster (unintended errors) and an intentional cluster (misdeeds). The base response for every type is to tell people what they need to know and do (instructing information) and to explain what happened, what is being corrected and that you care about those affected (adjusting information), and expressing concern is not an admission of guilt. Response strategies then scale with attributed responsibility: deny or correct for the victim cluster, diminish for accidents, rebuild with apology and correction for intentional harm or for an accident when there is a history of similar problems. Coombs also warns that over-apologizing for a crisis the audience does not blame you for can make them think it was worse than they assumed.

On timing, the relevant experiments are about stealing thunder, disclosing your own bad news before someone else does. Claeys, Cauberghe and Leysen's 2013 experiment in the Journal of Applied Communication Research, with 168 participants in a two-by-two design, found that ex-ante self-disclosure restored reputation better than ex-post response, that it made the framing of the message more effective, and that with ex-ante timing a spokesperson who expressed sadness produced a better post-crisis reputation than one who stayed purely rational. Lee's 2016 experiment in Public Relations Review found the benefit held when participants were not made aware of the organization's persuasive intent and disappeared when they were. These are small lab studies with students and consumers, and the effect sizes are not reported on the pages we could verify, so the protocol treats them as directional: if you have bad news about yourselves, be the first to say it, say it plainly, and do not dress it as spin.

Myth: you must respond to any attack within the "golden hour"

We found no peer-reviewed experiment supporting a one-hour rule for social media crisis response. The timing evidence that exists (Claeys and colleagues, 168 participants; Lee, 2016) is about disclosing your own problems before others reveal them, which is a different thing from racing to answer every critic. For someone else's false claim, the misinformation research in lesson 09 says what matters is that the correction carries the document, names the myth once, and reaches the audience that saw the claim. A careful correction at hour four beats a wrong one at minute twenty, and a rumor that is going nowhere is often best left alone.

How to: the crisis protocol on one page

  1. Classify within the first hour, not respond. Is this ours (we did something) or theirs (a false or hostile claim about us)? Who is affected? What is the attributed responsibility, in Coombs's terms?
  2. Ours: gather the facts, fix what can be fixed, and disclose first, on the page where the audience is, with the coordinator's or chair's name on it. Say what happened, what you are doing about it and that you regret the harm. Do not wait for the story to break. Do not hedge the apology if responsibility is yours.
  3. Theirs, false: verify against the document, then run the correction structure from lesson 09 (fact, myth once, technique, fact, document) in the same format and place as the claim. Do not quote-share the attack to your whole audience if it has little reach of its own.
  4. Theirs, hostile but true: treat as ours. The fact that an opponent found it does not change the response.
  5. Comments: one moderator on duty, bins from the moderation lesson, screenshots of threats and doxxing before removal, platform reports filed.
  6. Afterward: a note in the dashboard's Notes tab with dates, what was said, reach of the claim and of the response, and what the protocol should change. Every crisis improves the protocol or it was wasted.

Account security

Losing a page is the one operational failure a network cannot recover from in a campaign season, and it happens through the same two doors every time: a phished password and an unprotected admin. Google's security team, with researchers at New York University and the University of California, San Diego, measured how well basic protections stop hijacking and published the results on May 17, 2019. Adding a recovery phone number alone blocked up to 100 percent of automated bots, 99 percent of bulk phishing and 66 percent of targeted attacks. SMS codes blocked 100 percent of bots, 96 percent of bulk phishing and 76 percent of targeted attacks. On-device prompts blocked 100, 99 and 90 percent. The pattern is that any second factor stops almost all automated and mass attacks, and the stronger the factor the more it stops a determined attacker who knows your name. CISA's multifactor authentication guidance says the only widely available phishing-resistant method is FIDO/WebAuthn (hardware security keys and passkeys), urges organizations to move to it, suggests number matching to defeat push bombing where that is not yet possible, and states the ground rule: any MFA is better than none.

The site's privacy guide walks through setting these up on each platform. The network-level rules are below.

How to: lock down a page network

  1. Every person with any role on any account turns on two-factor authentication using an authenticator app, a passkey or a hardware key. SMS only as a fallback where the platform offers nothing else. Document the method next to each name in the role matrix.
  2. Two Admins on every asset, both with 2FA, neither sharing a login. Shared passwords are how a departing volunteer keeps access and how a phished password reaches everything at once.
  3. Unique passwords from a password manager for every account. The email account that receives password resets for the pages is the most important account you own; protect it with a hardware key.
  4. Phishing drill: show every admin what a fake "your Page will be unpublished" message looks like. Platforms do not message admins through Messenger or comments about policy violations; the real notices appear inside the account. Nobody clicks a link in an urgent message about the page. They open the app directly.
  5. Quarterly review: pull the people list on every account, remove anyone who left, downgrade anyone whose role changed, confirm 2FA is still on, check connected apps and revoke the ones nobody recognizes.
  6. Recovery kit: in a sealed envelope held by the chair, the backup codes for each account, the recovery email and phone for each, and the steps to recover. Test it once a year by having the chair log in.

Burnout and a cadence people can keep

Advocacy page work is volunteer or near-volunteer, emotionally charged, measured in public, and never finished. That is a burnout recipe. Maslach and Leiter's review in World Psychiatry in 2016 describes burnout as a prolonged response to chronic work stressors with three dimensions, exhaustion, cynicism and detachment, and a sense of ineffectiveness, and traces it to mismatches in six areas: workload, control, reward, community, fairness and values. They also note that intervention research is thin and rarely randomized, so what follows is the application of their framework, not a tested program.

Map the six areas onto a page team and the fixes are concrete. Workload: cadence set below the benchmark and held there, with the empty slot every week and a real rotation for comment moderation, which is the most draining task. Control: owners decide their own posts within the voice guide; the approval check is a checklist, not a committee. Reward: the dashboard shows actions and subscribers, so people see what their work did, and the thank-you posts the mobilization research recommends include the volunteers. Community: the weekly 20-minute review is in person or on a call, not a chat thread. Fairness: the role matrix is public to the team. Values: the ethics code is the team's own, and nobody is asked to post what they believe is false. A coordinator who can see exhaustion in the metrics (posts slipping, replies curt) acts on it as an operations problem.

Budgets by scale

Three scales cover most Idaho advocacy efforts. The figures below use prices and benchmarks verified in October 2026 and are annual. Email platform costs are shown from the two pricing pages verified for this lesson; lesson 07 compares the full set. Ad arithmetic uses M+R's nonprofit cost per lead on Meta ($3.64) and WordStream's traffic campaign cost per click ($0.77), with Triple Whale's $15.06 median CPM for scale, noting that the CPM comes from e-commerce advertisers and advocacy audiences in small markets can cost more or less.

LineVolunteer networkSmall nonprofitFunded campaign
Email platform$0 (Mailchimp Free to 250 contacts; Buttondown free to 100 subscribers) rising to about $13 to $20 a month at Mailchimp's introductory Essentials and Standard prices as the list grows$240 to $600 (paid tier sized to 2,500 to 5,000 subscribers; Buttondown offers 50 percent off for 501(c)(3)s)$1,200 to $3,000 (segmentation, automations, multiple senders)
Paid social: list building$0 to $500 (at $3.64 per lead, $500 is about 137 leads)$3,000 (about 824 leads at $3.64)$15,000 (about 4,100 leads)
Paid social: reach around hearings and votes$0 to $300$2,000 (at $15.06 CPM, about 133,000 impressions; at $0.77 CPC, about 2,600 clicks)$20,000 and up, with lift holdouts as in lesson 08
Design, video, tools$0 (free browser tools such as VideoStudio and ImageStudio)$600 (stock, a microphone, a tripod)$6,000 (contract videographer days)
Website and forms$0 to $150 (domain, free hosting tier)$400$2,000
Security (hardware keys for two admins)$60$120 (four keys)$300
Printing and QR signage for events$100$500$3,000
Staff timeVolunteer0.25 to 0.5 FTE1 to 2 FTE
Total cash (excluding staff)$160 to $1,110about $6,900 to $7,200about $47,000 and up

Two platform rules shape the paid lines. Google's political content policy does not allow ads related to ballot measures and candidates for local elections in Idaho, so Google and YouTube paid promotion is off the table for a levy or a commission race, though not for issue education outside an election context. TikTok's guidelines bar paid political promotion entirely, including paying creators to support or oppose a candidate. That leaves Meta as the main paid channel, with its authorization requirement for social issue, electoral and political ads and its Ad Library, which Meta's 2019 announcement said keeps political and issue ads for seven years. Lesson 06 covers the authorization process and the Idaho reporting that paid promotion triggers.

The worked 12-month plan

The network below is invented, as are its county, town and issues. Its numbers are not. Every rate comes from a benchmark cited in this course, and every milestone is a cell on the dashboard.

The Pine Bench Civic Network

Pine Bench County is a rural Idaho county of about 24,000 people with one town of size, Larkspur. Two issues are live over the coming year: a county proposal to restructure water rates for rural domestic wells, decided by the commission in a vote expected in month 8, and a road bond to replace the Larkspur Bridge, on the November ballot in month 11. A small group of residents has run a Facebook Page for two years (3,100 followers) and an email list of 400. They have formed an unincorporated association with a five-person steering group, one half-time coordinator funded by two local donors, and a budget of $7,200, the small nonprofit column above at its upper email figure.

Theory of change. The commission vote turns on three commissioners hearing from rural well owners, who are older, mostly on Facebook, and not on the email list. The bond needs a simple majority of a low-turnout electorate, and the persuadable voters are younger families in Larkspur who use the bridge daily and do not follow county politics. Per lesson 09, the commission objective is mobilization and agenda setting; the bond is a ballot measure, where information to the undecided has real effects.

Pages. Three, following the hub-and-spoke model from lesson 04: the existing main page renamed Pine Bench Civic Network (the voice and the hub); an issue page, Fix the Larkspur Bridge, run with a coalition of the volunteer fire department auxiliary and the school parents' group; and a chapter page, Ridgeline Well Owners, for the rural water issue, run by two residents of that area. Each page's About section names the network, its funding and the ethics code.

Platforms, justified. Facebook is primary because Pew's November 2025 report shows 57 percent of rural U.S. adults use it daily, the highest of any platform in rural areas, against 42 percent for YouTube, 25 percent for TikTok and 8 percent for X, and because daily use among adults 50 to 64 (54 percent) and 65 and over (45 percent) matches the well-owner audience. Instagram runs under the bridge page because 80 percent of adults 18 to 29 use it and the bond's undecided voters are young families. YouTube hosts the long explainers and hearing recordings, embedded on the site and in email, because 84 percent of adults use it and it is the archive the press can link. Email is the owned channel for every action. X is skipped: 21 percent of adults use it, 10 percent daily, and M+R's 2026 nonprofits saw their X followers fall 3 percent while 31 percent planned to leave. TikTok runs as a six-month test on the bridge page only, one video a week, because its nonprofit engagement rate (3.04 percent per video, Rival IQ 2025) and 34 percent follower growth (M+R 2026) are the highest of any platform and the bond's audience is young, with the knowledge that no paid promotion is possible there. Bluesky (4 percent of adults) and Nextdoor are left to the Ridgeline volunteers if they want them, with no targets attached.

The email goal. From 400 to 2,500 net subscribers in 12 months. Arithmetic: 2,100 net growth; churn at M+R's 2026 medians of 4 percent bounced and 12 percent unsubscribed, applied to the average list of about 1,450 over the year, costs about 230; gross signups needed are about 2,330, roughly 195 a month. Sources, planned: paid Meta lead ads $3,000 at $3.64 per lead gives about 824 (35 percent); website and content signups, 930 (40 percent), which at Unbounce's 6.6 percent median landing page conversion needs about 14,100 landing page sessions over the year, or about 1,175 a month, with paid social traffic converting nearer its 12 percent channel benchmark; events, QR codes and partner lists, about 575 (25 percent), from eight events averaging 72 signups each. Each source has its own UTM medium so the mix can be corrected quarterly.

QuarterOperationsContent and growthMilestones on the dashboard (end of quarter)
Q1 (months 1 to 3): foundationsBusiness Manager set up with two Admins and the role matrix; 2FA and hardware keys for admins; ethics code adopted and posted; voice guide and visual kit; calendar built; UTM convention written; dashboard live; email tool chosen from lesson 07's table, double opt-in and welcome sequence (3 emails) set; SPF, DKIM and DMARC configuredMain page at 3 posts a week, bridge and Ridgeline pages launched at 2 a week; first long YouTube explainer on how the water rate proposal works, with the county's own documents; newsletter every two weeks; 2 in-person listening events with QR signup; prebunking posts on the manipulation techniques both fights will see; no paid spend yet except a $300 lead-ad pilot to learn the real cost per leadList 400 to 760 net (gross about 400); pilot cost per lead recorded with a 95 percent interval; 3 pages with 2FA on every admin; approval checklist used on 100 percent of posts (status column); baseline engagement per reach per page recorded for 12 weeks (the A/A noise floor from lesson 08)
Q2 (months 4 to 6): build the well-owner audienceWeekly review running; first quarterly role matrix review; TikTok test begins on bridge page; first pre-registered experiment (video versus document carousel on the main page, pooled across 10 weeks)Ridgeline page carries personal stories of well owners (narrative evidence, lesson 05) and a monthly live Q&A; main page runs the rate-impact calculator post series with the arithmetic shown; lead ads $1,200 targeted to the county's rural zip codes with a tagged landing page; newsletter weekly from month 4; 3 events; agenda-setting push: op-ed submitted, commissioners asked on record whether rates are on the agendaList 760 to 1,400 net; cost per lead within 25 percent of the Q1 pilot; Ridgeline page reach among adults 50 and over (Insights age breakdown) above 50 percent of its reach; experiment result reported with denominators whether or not it moved; at least 2 mentions of the rate issue by commissioners or local press logged by hand
Q3 (months 7 to 9): the commission voteCrisis protocol rehearsed once with a tabletop scenario (a doctored chart circulates); comment moderation rotation doubled for the hearing weeks; approval check adds the electioneering-window test from month 9 onward as the bond nears its 60-day windowMobilization sequence for the month 8 hearing: three advocacy emails (M+R median response 1.4 percent of a 1,700 list is about 24 actions per send; the plan asks for 120 confirmed speakers and letters across the sequence plus Facebook event RSVPs); public thank-you posts naming attendees who consent (social accountability, Gerber, Green and Larimer); $800 of boosted posts to rural zip codes around the hearing with a two-zip-code holdout; after the vote, a plain results post and a correction log reviewList 1,400 to 1,950 net; 120 or more documented hearing actions (sign-in sheet reconciled against tagged clicks); boosted-versus-holdout zip codes compared and recorded; zero posts failing the pre-publish check; TikTok test decision made on its own pre-registered criteria (reach among 18 to 29 and tagged clicks per video)
Q4 (months 10 to 12): the bond and the handoffTreasurer confirms whether any spend constitutes an independent expenditure or electioneering communication and files per the campaign finance course; all express advocacy carries the 67-6614A identification; synthetic media disclosure rule reiterated to all admins; year-end role review; recovery kit testedBridge page leads: short videos of the bridge, a cost-per-household table, a Q&A on what the bond does and does not fund (prebunked against the trail rumor); Instagram and TikTok carry the young-family stories; lead ads $1,500 to Larkspur zips, bringing lead-ad spend to the $3,000 budgeted; $1,200 of boosted posts to Larkspur zips in the final three weeks with a holdout pair, bringing reach spend to the $2,000 budgeted; two advocacy emails on voting dates with tagged links; one fundraising email; post-election: results, thanks, what the network does next, and the annual results memo from lesson 08List 1,950 to 2,500 net; bridge page tagged clicks per post above Q3 median; Larkspur share of email list (by zip at signup) above 40 percent; annual memo delivered with every objective's primary metric, denominators, tests and the stopped list; budget closed within 5 percent of $7,200

The plan does not promise the commission vote or the bond. The persuasion research says a page cannot promise those. It promises what the page can control and the board can verify: an audience built where the decision makers' constituents actually are, an owned list six times larger than it started, documented actions at the hearing, every rule followed, and an honest account of what did and did not work. If the Ridgeline well owners fill the hearing room and the commissioners feel it, that is the mechanism the evidence supports. If the bond passes, the undecided young families who learned what it funded from a labeled, documented, plain-language video are the voters the ballot measure research says can be moved.

If you report on this

  • A network's About sections should name the organization, funding and admins. Compare them across the pages; a page that hides its connection to the others is the story.
  • Search the Meta Ad Library for each page name and for the organization; political and issue ads are retained for seven years with spend and impression ranges. Reconcile the spend against the Sunshine portal filings the campaign finance course describes.
  • Ask for the ethics code, the corrections log and the role matrix. A network that cannot say who holds admin on its pages has an accountability problem before it has a message problem.
  • When a page claims a hearing turnout, ask for the sign-in sheet count, not the Facebook RSVP count, and ask how the two were reconciled.

Go further

Questions and answers

We have one volunteer. Which of the four documents matters most?

The role matrix, even if it has two names on it, because it forces the second admin and the 2FA that keep the page alive. Then the pre-publish checklist, because the legal mistakes in lesson 09 are made by tired people at night. The calendar and crisis protocol can be a page each.

Our coordinator wants to post daily on every page. Is more always better?

The benchmarks say mid-sized nonprofits post 4 to 6 times a week on Facebook, and the evidence in lesson 05 on frequency is weaker than the folklore. For a team that will still exist in month nine, a cadence you can keep beats one you cannot. Start at the volunteer column, hold it for a quarter, read the dashboard, then decide.

Why hardware keys when the Google numbers for SMS are already high?

Because the gap is in targeted attacks: SMS blocked 76 percent, on-device prompts 90 percent, and CISA calls FIDO/WebAuthn the only widely available phishing-resistant option. A network that opposes a funded interest in an election year is a target, not a random victim. Two keys cost about $60.

Should we announce our own mistake if nobody has noticed yet?

The stealing thunder experiments say disclosing first protects reputation better than being found out, with the caveat that the studies are small and the benefit faded when the audience saw the disclosure as spin. If the error affected what people believe about a public matter, correct it and say so. If it was a typo, fix it and move on.

How do we pick which zip codes get the boost and which are the holdout?

Pair them on population and past engagement, flip a coin, and write the assignment in the pre-registration note before spending. Swap roles for the next campaign. Compare tagged signups or RSVPs per 1,000 residents between the pair. It is crude, and it is still the only design that tells you whether the money did anything, as lesson 08 explains.

The plan skips X. What if our county's reporters live there?

Then one person posts links there without targets or cadence, and you measure the tagged clicks. The plan skips X as a growth channel because of its 10 percent daily use and the nonprofit follower declines in M+R's data, not because reaching three reporters is worthless. The decision table in lesson 03 is built for exactly this trade-off.

What does the plan do if cost per lead comes in at $9 instead of $3.64?

That is what the Q1 pilot is for. At $9 the $3,000 buys about 333 leads instead of 824, so the plan either raises the event and website share of signups or accepts a lower list target and says so in the Q1 memo. M+R's health sector median was $9.25, so the number is not implausible for a narrow rural audience. Adjusting the plan in public is the whole point of having one.

Do we need the full ethics code before launch?

Yes. It takes an hour, it protects the volunteers who follow it, and the first opponent who finds an undisclosed connection or an unlabeled image will ask whether you have one. The template is in lesson 09.

Hard questions

Isn't a 12-month plan fantasy for volunteers who do not know what next month looks like?

The plan is a set of defaults and milestones, not a prediction. Its first quarter is almost entirely setup that any network needs, and its growth arithmetic is built from medians with a pilot to correct the one number that matters most. What a plan buys a volunteer group is the ability to say no: no to the fourth platform, no to the daily post, no to the unbudgeted boost. And it gives the board a dashboard cell to check instead of a feeling.

The benchmarks come from national charities and e-commerce brands. Why trust them for a rural Idaho group?

Do not trust them; use them as starting points and replace them with your own numbers by the end of Q1. The lesson says where each figure comes from and what population produced it: Rival IQ's are mid-sized organizations with 25,000-plus Facebook fans, M+R's are 180 nonprofits with online fundraising programs, Triple Whale's CPM is e-commerce. The alternative to a sourced starting point is an unsourced one. The pilot and the dashboard are how the plan stops depending on anyone else's data.

Why spend $5,000 on Meta ads when the persuasion research says ads barely move anyone?

Because the plan's ads are not trying to persuade. $3,000 buys contact information from people who already care enough to give it, and $2,000 puts hearing dates and voting information in front of likely supporters, which is mobilization and reach. The small-effects research is about changing candidate preference with ad content. Coppock, Hill and Vavreck's 0.05-point favorability effect is irrelevant to whether a lead ad produces a subscriber at $3.64, which is a question the dashboard answers in a month.

Doesn't all this process kill the spontaneity that makes a page worth following?

The process is a three-minute checklist, a shared calendar with an empty slot every week, and a weekly 20-minute call. The voice guide exists so volunteers can write freely inside known lines rather than asking permission for every post. What kills pages is the opposite: the unlabeled image that becomes a story, the lost admin login, the coordinator who burns out. Maslach and Leiter's six mismatches are mostly about control and fairness, and clear roles give people more of both.

Why should a board care about any of this instead of the vote count?

Because the vote count is the one thing the research says the page cannot promise, and a board that judges the page by it will either fire a good team after a loss or keep a bad one after a win. The milestones in the plan are the things a page controls and the mechanisms the evidence links to outcomes: audience built where the constituents are, documented actions, an owned list, rules followed. A board that checks those is holding the team to a standard it can meet and learn from.

Pictures, Numbers and Infographics

The figures below are the ones the plan rests on, each with its source, population and date. Where a figure is industry-sourced or from a sample unlike a small advocacy group, the table says so. All checked October 2026.

Platform use by age and community, U.S. adults, 2025

PlatformAll adults ever useDaily use, 18 to 29Daily use, 30 to 49Daily use, 50 to 64Daily use, 65+Daily use, ruralDaily use, suburbanDaily use, urban
YouTube84%66%54%45%27%42%48%55%
Facebook71%49%58%54%45%57%51%49%
Instagram50% (80% of 18 to 29; 19% of 65+)Daily use by age and community not reported in the sources verified
TikTok37%47%28%17%5%25%21%28%
Reddit26%Not reported by age in the sources verified
X21%18%10%9%4%8%11%10%
Threads8%Not reported by age in the sources verified
Bluesky4%Not reported by age in the sources verified

Source: Pew Research Center. The "ever use" column is from the National Public Opinion Reference Survey, February 5 to June 18, 2025, 5,022 U.S. adults, address-based probability sample (Social Media Fact Sheet). The daily-use columns are from a separate Pew survey of 5,123 U.S. adults conducted February 24 to March 2, 2025, reported in the appendix of Americans' Social Media Use 2025 (Nov. 20, 2025); "daily" combines "several times a day" and "about once a day." National figures; Idaho-specific breakdowns are not published.

Engagement and posting benchmarks, 2024 data

PlatformAll-industry median engagement rate per post (by follower)Nonprofit median engagement rate per postAll-industry median posts per weekNonprofit median posts per weekNonprofit follower growth 2025 (M+R)
Facebook0.046%0.046%4.195.51+3%
Instagram0.36%0.561%4.54.18Not reported
TikTok1.73%3.04%1.971.42+34% (37% in the chapter text; the page is inconsistent)
X0.015%0.023%2.165.68-3%

Sources: Rival IQ 2025 Social Media Industry Benchmark Report (150 companies per industry sampled from 200,000, Facebook fan counts 25,000 to 1,000,000, 2024 data, medians; engagement rate is total interactions divided by followers); M+R Benchmarks 2026, Social Media (180 nonprofits, 2025 data). Rival IQ's sample is larger organizations than most advocacy networks; expect your rates to differ.

Email benchmarks

MetricFigurePopulation and dateCaveat
Nonprofit open rate40.04% (Mailchimp); 26.6% (Campaign Monitor)Mailchimp campaigns to 1,000+ subscribers, Dec. 2023; Campaign Monitor, 2021 sendsBoth sources warn Apple Mail Privacy Protection inflates opens; do not plan on opens
Nonprofit click rate3.27% (Mailchimp); 2.7% (Campaign Monitor)SameIndustry platforms, all nonprofit mail mixed
Government and politics click rate2.8% (Campaign Monitor); government 4.58% (Mailchimp FAQ)SameCategories differ between the two
Nonprofit unsubscribe rate per send0.18% (Mailchimp); 0.2% (Campaign Monitor)Same
Advocacy email click-through rate2.3%M+R 2026, 180 nonprofits, 2025 sendsMedians of organizations with online programs
Advocacy email response rate (took the action)1.4%M+R 2026About 14 actions per 1,000 recipients
Fundraising email click-through / response rate0.59% / 0.05%M+R 2026About $54 per 1,000 fundraising emails
Donation after an advocacy action0.34%M+R 2026
Annual list churn4% bounced away plus 12% unsubscribedM+R 2026Of starting subscribers
Annual list growth5% net medianM+R 2026The plan's 525 percent is a startup number, not a steady state
Email share of online revenue11%M+R 2026
Landing page conversion, email traffic19.3%Unbounce 2024, 41,000 pagesCommercial landing pages; highest of any channel

Sources: Mailchimp, Campaign Monitor, M+R Benchmarks 2026, Unbounce 2024.

Ad cost and conversion benchmarks

MetricFigurePopulation and dateCaveat
Meta ads median CPM$15.06Triple Whale, 40,000+ e-commerce brands, Aug. 2025 to Jul. 2026E-commerce, not advocacy; use for scale only
Facebook traffic campaigns: CTR / CPC1.57% / $0.77WordStream, 2,946 campaigns, Feb. 2023 to Apr. 2024, mediansBusiness advertisers
Facebook lead campaigns: CTR / CPC / conversion / cost per lead2.53% / $1.88 / 8.78% / $21.98WordStream, sameBusiness leads; nonprofit leads are far cheaper below
Nonprofit cost per lead, Meta / all channels$3.64 / $2.81M+R 2026Health sector $9.25; display $86
Nonprofit return on ad spend, search$2.48 per $1M+R 2026Fundraising benchmark; TikTok $0.04, digital video $0.56
Nonprofit cost per donation, TikTok$590M+R 2026TikTok also bars paid political promotion
Nonprofit digital ad spend per $1 raised online$0.10M+R 2026Environmental $0.20, health $0.21
Landing page conversion, all industries median6.6%Unbounce 2024Paid social traffic 12%; Facebook 13%; Instagram 17.9%
Donation page conversion (all site visitors)1.6%M+R 2026Main donation page: 11% desktop, 8% mobile
Google ads for Idaho local elections and ballot measuresNot allowedGoogle political content policy, Oct. 2026Issue education outside elections is separate
TikTok paid political promotionNot allowedTikTok Community Guidelines, Oct. 2026Includes paying creators

Sources: Triple Whale, WordStream/LocaliQ, M+R Benchmarks 2026, Advertising and Website Performance, Unbounce 2024, Google Ads policy, TikTok.

The social-to-email funnel with benchmark rates

100,000 impressions (paid social) at $15.06 CPM: about $1,506 (Triple Whale 2025 to 2026) 1,570 link clicks 1.57% click-through, traffic campaigns (WordStream 2024) 188 signups (104 at the 6.6% all-channel median) 12% landing page conversion from paid social (Unbounce 2024) 158 still on the list after a year 4% bounce plus 12% unsubscribe churn (M+R 2026) about 2 actions per advocacy email 1.4% response rate per send (M+R 2026) about 1 donation per 300 actions 0.34% post-action donation (M+R 2026) Cost per signup from this funnel: about $8 at 12% conversion, about $14.50 at 6.6%. M+R's nonprofit median cost per Meta lead ($3.64) uses lead forms inside the platform, which skip the landing page step.
The social-to-email funnel for 100,000 paid impressions, using the benchmark rate at each step. Every figure is from the tables above; the plan's own numbers replace them after the first quarter.

Read the funnel from the bottom. Two actions per advocacy email from a cohort of 158 is what a $1,500 spend buys in direct response, which is why the plan treats paid impressions as a way to acquire a list and treats the list, over many sends and many months, as the way to get action. The same $1,500 spent on in-platform lead forms at M+R's $3.64 median produces about 410 subscribers, which is the arithmetic behind the plan's choice to put most of its paid budget into lead ads rather than traffic. Whether that holds in a rural Idaho county is the Q1 pilot's job to find out.

Check yourself

1. Google's 2019 research with NYU and UCSD measured how well account protections blocked hijacking. What did SMS codes sent to a recovery phone block?

2. In Meta's Business Manager, what can an Employee role do by default?

3. Claeys, Cauberghe and Leysen's 2013 experiment with 168 participants compared disclosing a crisis before others did (ex-ante) with responding after. What did they find?

4. The worked plan sets a net email list target of 2,500 from a start of 400. Using M+R's 2026 churn figures (4 percent bounced away, 12 percent unsubscribed), roughly how many gross signups does the year need?

5. Why does the worked plan run Facebook as the primary platform for a rural Idaho network?

6. What does M+R's 2026 benchmark report give as the median cost per lead for nonprofit Meta lead generation ads?

Terms in this lesson

Editorial calendar
A dated schedule of what each page will publish, by format and objective, with owners and approval status.
Role matrix
A table of who holds which permission on which page and account, reviewed on a fixed schedule.
Business Manager
Meta's tool for holding Pages, ad accounts and other assets under one organization with Admin and Employee roles and per-asset tasks.
Stealing thunder
Disclosing bad news about your own organization before others reveal it; experiments find it protects reputation better than responding afterward.
Situational Crisis Communication Theory
Coombs's framework matching the response (deny, diminish, rebuild, bolster) to how much responsibility the audience attributes to the organization.
Phishing-resistant MFA
Two-factor methods such as FIDO/WebAuthn security keys that cannot be captured by a fake login page; CISA's preferred form.
Burnout
Exhaustion, cynicism and a sense of ineffectiveness arising from chronic work stress, tied to mismatches in workload, control, reward, community, fairness and values.
Cost per lead
Ad spend divided by the number of people who gave contact information; the unit for budgeting list growth from paid social.
Churn
The share of an email list lost in a period to bounces and unsubscribes; M+R's 2026 median is 4 percent plus 12 percent.
Milestone
A dated, measurable target in the plan that a board can check against the dashboard without interpretation.

Full glossary

Sources