Concept paper · Playbook · Operations manual

Trusted Voices: DHA Digital Workforce and Creator Strategy

Concept paper, playbook and operations manual · Working draft for the DHA digital strategy working group

15 October 2026 · Andrew D. Plummer

Executive summary

DHA should build a governed digital workforce: trusted people, inside the agency first, who tell the story of military medicine on the platforms where beneficiaries, staff and partners already spend their attention, paired with sentiment measurement that turns audience reaction into a standing input to product and service decisions.

The case is simple. The Military Health System has a message-delivery problem and a feedback problem, and they are the same problem. Newsletters, web pages and post-rollout surveys reach a shrinking share of a population that is mostly under 35 and gets its information from people it trusts, not from institutions. MyOp showed the cost: the audience told DHA what it thought, loudly and in public, and DHA learned it late and passively.

The proposal has three parts that move together:

1. A trust layer: vetted clinicians, researchers, interns and, later, external creators speaking in their own voice with accurate talking points.

2. A sensing layer: social listening and sentiment analysis that reports how messages, products and services are landing, continuously rather than after the fact.

3. A governance spine: creator vetting, message alignment with the Department, and a crawl-walk-run gate structure so leadership approves a bounded proof of concept rather than a philosophy.

The external evidence is strong. The Top-30 military and veteran creator map (counts as of 15 Oct 2026) finds roughly 98.5M combined followers across Instagram, TikTok, YouTube and X, with the top five creators holding about 48% of it, 13 of 30 already tied to DoD or another government body, and active-duty short-video creators reaching 100x or more their follower counts. No healthcare organization has run this play well at scale. Under the Director's intent to lead the healthcare industry in the digital space, that gap is the opportunity, not the objection.

Leadership is asked to approve one bounded proof of concept (Section 11), the governance standard that bounds it (Section 7), and early engagement with requirements and acquisition so the model can scale if it works.

The problem

DHA cannot reliably reach its audiences, and it cannot reliably hear them. Both failures have the same cause: the agency communicates through channels its audiences stopped using as primary sources, and it measures reaction through instruments (post-rollout surveys, media monitoring) that report weeks late and at low resolution.

The information environment moved. Trust has shifted from institutions toward individuals. The 2025 Edelman Trust Barometer puts scientists (77) and "my employer" (75) well above government (52) and media (52) as trusted sources, and its 2024 workplace report found employees more likely to name co-workers than executives as trusted sources (Edelman 2025). Younger audiences validate what institutions say by checking what people they follow say. A newsletter, a glossy or a web page is a claim; a clinician on camera is evidence.

The MHS audience is young and mobile. Most service members are under 35, and their families and younger beneficiaries consume information through TikTok, YouTube Shorts, Instagram Reels and Snapchat, with LinkedIn as the professional channel. Retirees still use social media, mostly Facebook. The proposed MHS monthly newsletter is good content with no distribution.

DHA learns late and passively. MyOp is the case in point. The audience reported what it thought of the rollout in public, in volume, in near real time. DHA observed that signal rather than using it. Today the agency has no routine way to know whether a capability, a service (urgent care access, telehealth, the Nurse Advice Line) or a message is landing with end users until a survey, a complaint pattern or a news story surfaces it.

Nobody has solved this in healthcare, for good reasons. Healthcare and defense are the two sectors where this is rarely done: HIPAA and patient privacy, conservative public affairs posture, and the need for message alignment with the Department all make organizations reluctant to let authentic voices show blemishes. The proposal does not argue those constraints away. It argues that a governed program can satisfy them, and that the Director's intent to lead the healthcare industry in the digital space makes the gap a first-mover opportunity.

What this is not. It is not a plan to hire influencers to say nice things. It is not a replacement for DCPA channels. It is a deliberate, consistent process for telling the story through people the audience already trusts, and for treating what the audience says back as data.

The concept: a feedback loop with people in it

The program is three layers that only work together. Remove the trust layer and DHA is just listening. Remove the sensing layer and DHA is just broadcasting through nicer channels. Remove the governance spine and leadership will not, and should not, approve either.

Audiences. Four external buckets from the working deck, plus one the group agreed to add explicitly: military families and veterans; beneficiaries; healthcare professionals inside and outside the MHS; the science, technology and research community; the fitness and wellness community; and DHA's own workforce, which is both an audience and the first cohort of voices.

Strategic objectives. Build trust, accelerate adoption of DHA messages, products and initiatives, and improve readiness. The measure of success is whether the loop changes a decision: a product optimized, a service fixed, a rumor corrected, a rollout adjusted before the survey would have caught it.

Two directions of use. The group identified both, and the proof of concept should pick one. Push: drive end-user adoption of a capability DHA has fielded. Pull: find out whether beneficiaries know about and can reach a service they already have (urgent care access, telehealth, the Nurse Advice Line). Pull is cheaper to measure and more likely to surface what is broken.

Why start inside. Internal voices carry the highest trust with leadership, are already bound by DoD rules, and produce the proof points that make external partnerships approvable. Mount Sinai built a 320-ambassador physician network before it ever paid an external creator (Section 5). The Army tried the reverse order and paused its pilot for legal review.

Evidence base: the Top-30 military and veteran creator map

The external creator pool is large, concentrated, already government-adjacent, and rewards format and authenticity over follower count. Those four facts drive the design choices in Sections 8 and 11. Counts are as of 15 Oct 2026 from Social Blade and public profiles; the full dataset with a link for every value is in Military_Influencer_Map_2026.xlsx.

Caveats that travel with the numbers. Reach figures are self- or press-reported and unaudited. The 98.5M total is a footprint, not unique people. Sponsor detections from third-party tools are unconfirmed. Niche labels are the author's judgment. One creator was excluded over a reported 2026 indictment; the charges are allegations. The set covers four platforms and omits LinkedIn, which is the channel for the professional and federal-health-IT audience (see Section 8, crawl phase).

A companion interactive page, Military Creator Pearls, carries the sortable leaderboard, the engagement and reach charts and the source list for the working group.

Who does this exceptionally well

The strongest models share three traits: they started with their own people, they wired listening into operations rather than into a report, and they published the rules before they published content. Every entry below was verified on the linked page on 15 Oct 2026; results are as the organization or its evaluator reported them.

Government and military

Healthcare

Service industries

Employee advocacy and listening at enterprise scale

What could not be verified and was left out. An Air Force Recruiting Service contract with Sam Eckholm (a 2026 Air Force and Space Force series is confirmed; the recruiting-command tie is not). A VA creator program. A CDC-run influencer program (only CDC-supported partner pilots are documented). Employee-creator elements at Chipotle. Membership totals for Mayo's network.

Best practices

Twelve principles, each traceable to a case in Section 5 or a finding in Section 4.

1. Start with your own people. Mount Sinai, NYC Health + Hospitals, Dell and NASA all built on employees before they paid outsiders. Internal voices are already bound by the rules, carry the highest leadership trust, and generate the proof points that make external work approvable.

2. Nominate, train, certify, then release. Mount Sinai's chairs nominated the first 20 physicians; Dell certified 15,000 employees through tiered coursework before letting them speak for the company. Training is the governance control that scales.

3. Pay in access, not cash. NASA Socials and the Army Creative Reserve run on event access and travel. The Top-30 creators monetize through owned brands, so a sponsorship pitch competes with their own products. Access also avoids the monetization problem that stalled the Army pilot.

4. Rank voices by reach and engagement, not followers. An active-duty creator with 300K followers reported 35-45M monthly reach; a 6M-follower account sits at 0.04% engagement; Marriott's mid-tier creators out-engaged its celebrities. Follower count is the wrong sort key.

5. Pair the expert with the creator. HHS paired physicians with celebrities on Instagram Live; the Coast Guard paired a public affairs specialist with a science YouTuber. The expert carries accuracy; the creator carries the audience.

6. Let authentic voices show the blemish, and fix it. A DHA employee who criticized the agency on TikTok and then posted that DHA fixed it would carry more weight than any good-news story. Dell's command center converted a third of "ranters to ravers" by resolving cases in public. Authenticity is only an asset if the organization closes the loop.

7. Wire listening to operations, not to a report. Southwest put a listening desk inside its network operations center. The MyOp signal was available in real time; nobody was positioned to act on it. The sensing layer earns its budget when it changes a decision inside the same week.

8. Publish the rules before the content. The UK GCS guidelines set tiers, due diligence, labelling, sign-off and transparency reporting up front. Every program that paused did so over a rule it had not written down.

9. Vet like it is a security investigation. GCS reviews 12 months of a creator's content and runs conflict checks before engagement. The creator map's one exclusion, over a reported indictment, is the reminder that a screenshot from years ago is a program-ending event. AI-assisted review of a creator's history is a reasonable accelerator; the decision stays human.

10. Design attribution before launch. HHS, NACHC and Philadelphia all reported strong engagement and no attributable outcome because outcome tracking was not built in. Marriott got a lift study because it planned one. The proof of concept must name its outcome metric and its measurement method on day one.

11. Keep the team small and fast. Duolingo runs one of the largest brand accounts with three people and a six-day cycle. A creator who waits two weeks for approval of a 30-second video is a creator who stops posting.

12. Own the editorial strategy centrally. Cleveland Clinic's content team, not its departments, decides what gets made. DCPA should own the editorial calendar and the talking-point library; directorates feed it, they do not program it.

Governance and guardrails

The rules that bound this program already exist; the program's job is to apply them, not invent them. The binding constraint is monetization: DoD personnel may not use their status, name, image or likeness for private gain, and that rule is now punitive. An unpaid, access-based internal program sits comfortably inside it. A paid serving-member program does not, which is why the Army's paused.

The rulebook, verified 15 Oct 2026

No DHA-specific social media instruction is public; the health.mil policy library returns nothing for the query, and the MHS social media page defers to the DoD issuances above. The program should either confirm an internal DHA instruction exists or write one as a crawl-phase deliverable.

Program controls

1. Vetting standard. Adapted from the UK GCS: 12 months of public content reviewed, conflict-of-interest and prior-affiliation check, OPSEC review, and for external creators a search of enforcement and litigation records. AI-assisted review may triage; a named human signs the clearance. Re-vet annually and on any incident.

2. Talking-point library and alignment. DCPA maintains approved themes and facts; the Department-level alignment question (what the ceiling for vetting is, who provides top cover) is settled in writing with OATSD(PA) before the walk phase. Creators get facts and themes, never scripts.

3. Disclosure. Every program participant discloses the relationship in the content itself ("I work at DHA"; "DHA hosted me"). Personal accounts carry the DoDI 5400.17 disclaimer. No seals, no official photos on personal accounts.

4. Compensation. None for DoD personnel beyond travel and access. External creators may be paid only through a contract vehicle, with FTC-standard disclosure, and only in the run phase.

5. Content review. Pre-publication review for internal voices is limited to accuracy, OPSEC, privacy and ethics; it does not edit voice. Target turnaround: 48 hours. Content that touches a policy under deliberation, a litigated matter or patient care escalates.

6. Escalation and incident path. A single named owner in DCPA; OGC ethics counsel on call; a 24-hour standard for a first response to a public complaint surfaced by the sensing layer; a stop-work authority that can pause any participant or the program.

7. Records. Official-account content and sensing-layer outputs are federal records under the OSD archiving procedures linked from health.mil; personal-account content is not, and the program does not attempt to control it beyond the rules above.

Playbook: crawl, walk, run

The program advances only through named gates, each with criteria leadership can check. Calendar dates are deliberately absent; the working deck's dates were placeholders, and the gates, not the calendar, decide when a phase ends.

The crawl phase is the accent because it is the only phase being proposed for approval now; walk and run are described so leadership can see where it leads.

Crawl: internal digital workforce

• Objective: establish the process and governance, validate the model with people DHA already trusts, and produce one proof point leadership can count.

• Voices: DHA clinicians, researchers, interns and staff, nominated by directorate leads and volunteered, never assigned. Target cohort of 15-25, modeled on Mount Sinai's launch group of 20.

• Entry criteria: working-group charter signed; DCPA named as editorial owner; ethics counsel assigned; vetting standard and disclosure wording approved; sensing-layer tool or interim method selected; proof-of-concept target and metric named (Section 11).

• Activities: cohort training (DoDI 5400.17 Section 8, JER 2-508, 5 CFR 3601.105 disclaimer, HIPAA, OPSEC, platform craft); talking-point library stood up; 48-hour accuracy review in place; weekly listening report to the working group; a LinkedIn Live or roundtable on a DHA topic with a platform or industry partner, which DCPA has standing relationships to arrange; a compilation of residents and clinicians already posting about military medicine, built with Anjali's team.

• Exit criteria (Gate 1): proof-of-concept metric met or a documented reason it was not; zero ethics or OPSEC incidents, or incidents closed to counsel's satisfaction; leadership dashboard delivered and reviewed at least twice; written Department-level alignment on message vetting authority.

Walk: DHA's existing creator community

• Objective: extend the model to serving and veteran creators who already tell the military-medicine story, without paying them.

• Voices: the creators Anjali's team has identified, plus the government-tied names in the creator map (Section 4). Engagement is access-based: embeds, interviews with subject-matter experts, early briefings on rollouts, event invitations.

• Entry criteria: Gate 1 passed; vetting standard proven on the internal cohort; alignment with the Department settled in writing; requirements and acquisition engaged on the sensing-layer vendor.

• Activities: creator briefings tied to specific rollouts; a response network able to correct misinformation within 24 hours of the sensing layer flagging it; the listening desk formally connected to the operational owners of the products and services it monitors; second-cohort internal onboarding.

• Exit criteria (Gate 2): at least one documented instance where the loop changed an operational decision; sentiment and reach trend reported for a full quarter; vendor contract or interim capability in place; no unresolved alignment dispute.

Run: external creator network

• Objective: reach growth-market audiences disconnected from DHA through creators in medicine, science, technology, fitness and wellness.

• Voices: external creators selected by niche and engagement rate, not follower count, from the creator map and DCPA's list.

• Entry criteria: Gate 2 passed; a contract vehicle for paid creator work with FTC-standard disclosure; a funded sensing-layer capability; leadership approval of the risk appetite for paid external voices.

• Activities: campaign-based engagements with lift measurement (the Marriott model); cohort structure with selection and mentorship (the Sephora model); annual re-vetting; transparency reporting of spend, following the UK GCS practice.

• Exit criteria: this phase does not end; it is reviewed annually against the measurement framework.

Operations manual

The program runs on one accountable owner, a weekly rhythm, five standing workflows and four checklists. Everything below is sized for the crawl phase and scales by adding people to roles, not by adding roles.

Roles and responsibilities

Battle rhythm

Workflow 1: identify and vet

1. Nominate (internal) or source (external) a candidate; creator liaison opens a vetting packet.

2. Review 12 months of public content on every platform the candidate uses; run conflict-of-interest, prior-affiliation and OPSEC checks; for external candidates, search enforcement and litigation records. AI-assisted triage may surface items; a named reviewer reads them.

3. Ethics counsel signs the clearance or returns it with reasons. Target: 10 working days.

4. Record the clearance, its date and its reviewer; set the annual re-vet date.

Workflow 2: onboard and brief

1. Participant completes the training module (DoDI 5400.17 Section 8, JER 2-508, the 5 CFR 3601.105 disclaimer, HIPAA and patient-image rules, OPSEC, platform craft).

2. Participant signs the program agreement: disclosure wording, no compensation beyond travel and access, escalation contacts, right to withdraw at any time.

3. Creator liaison delivers the current talking-point library and the editorial calendar; the participant chooses what to make.

Workflow 3: content pipeline and review

1. Participant drafts content in their own voice and submits it for review when it touches an official program, a rollout, a statistic or a policy. Personal content that does not touch official matters is not reviewed.

2. Reviewer checks accuracy, OPSEC, privacy and ethics only, within 48 hours. Voice is not edited.

3. Content that touches a policy under deliberation, a litigated matter or patient care escalates to the program owner and ethics counsel.

4. Published content is logged with its URL, date, participant and topic so the sensing layer can attribute reaction to it.

Workflow 4: listen and respond

1. The sensing layer monitors named products, services, messages and the program's own content across the platforms in scope.

2. A flag (sentiment shift, misinformation, service complaint pattern) is routed to the product owner and the program owner the same day.

3. The product owner decides the operational response; the program owner decides whether a trusted voice should address it publicly. A first public response, where one is warranted, lands within 24 hours.

4. Every flag is closed with a disposition: acted on, monitored, or no action, with a reason. Closed flags are the raw material for the "loop changed a decision" metric.

Workflow 5: incident handling

1. Any participant, reviewer or listener can raise an incident: an ethics question, an OPSEC concern, a privacy breach, a misstatement, or a participant's unrelated conduct becoming news.

2. The program owner pauses the affected content or participant within hours; ethics counsel and the executive sponsor are notified the same day.

3. Counsel reviews; the program owner drafts any public correction, which the participant delivers in their own voice where appropriate.

4. The incident is closed with a written finding and any change to the standard. The program is paused as a whole only by the executive sponsor.

Checklists

• Vetting packet complete: 12-month content review, conflict check, OPSEC check, enforcement search (external), reviewer signature, re-vet date

• Onboarding complete: training, signed agreement, talking-point library delivered, escalation contacts confirmed

• Content ready: disclosure present, disclaimer present where title or rank is used, no seals or official photos on personal accounts, no patient information or images, facts traceable to the library, OPSEC cleared

• Flag closed: product owner notified, disposition recorded, public response logged if any, dashboard updated

Measurement framework and leadership dashboard

The program is measured on outcomes, not impressions, using the AMEC Integrated Evaluation Framework, the non-proprietary standard 2,000+ organizations use and the one that operationalizes the Barcelona Principles (now at version 4.0, which bans advertising-value equivalents and treats audiences as active participants). The framework's chain is objectives, inputs, activities, outputs, out-takes, outcomes, impact. The five measures in the working deck map onto it as follows.

The single number leadership should ask for is decisions changed: the count of closed listening flags whose disposition was "acted on," with the action named. It is the only measure that proves the loop exists. Everything else proves the loop is running.

Dashboard. One page, refreshed weekly, built by Data and Analytics with the performance lead's definitions. Top row: the proof-of-concept metric against its target, decisions changed this month, net sentiment for the monitored services. Middle: reach and engagement by voice and by platform, with the cohort's engagement rate against the creator-map median of 1.02% as a benchmark. Bottom: open flags by age, time-to-first-response, incidents. The deck's sample dashboard and its backup version are the starting layouts; the performance lead chooses between them.

Attribution. Platform-reported reach is a footprint, not people, and reach claims from creators are unaudited. For the proof of concept, attribution rests on three things set before launch: a baseline period of at least four weeks for the targeted service's utilization and sentiment; tagged links or a named referral path from program content to the service; and a comparison group where one exists (a service or region not covered by program content). A platform brand-lift study, as Marriott used, is an option once there is a paid-media budget, which is a run-phase question.

What the sensing layer must do to be useful. Monitor named entities (DHA, MHS, TRICARE, MyOp, the targeted service) across the platforms in scope; separate beneficiary, staff and public voices where the platform allows; surface a flag with an example post and a volume trend, not a score alone; and export weekly so the analytics lead can code a sample by hand. DoDI 5400.17 Section 7.2 authorizes this use of publicly available information, and DoDI 5400.19 governs any AI inside it.

Proof of concept proposal

Recommend one bounded target: beneficiary awareness and use of a service DHA already provides, in one market, over 90 days, told by DHA's own people. The working group's own example, knowing that urgent care access or a telehealth appointment is available rather than going to a civilian urgent care, is the right shape. It is a pull problem (do people know and can they reach it), it has a utilization number the product owner already tracks, it carries no policy controversy, and it is the kind of service where the listening layer will surface what is broken quickly.

Candidate targets for the working group to choose from

Success criteria, set before launch and not revised after

1. Outcome: a measurable change in the targeted service's utilization or awareness against a four-week baseline, with a comparison market where one exists. The performance lead sets the threshold with the product owner; it is written into the charter.

2. Loop: at least three listening flags closed as "acted on" by the product owner, with the action named.

3. Reach and engagement: cohort content reaches a stated share of the target market and sustains an engagement rate above the creator-map median of 1.02%.

4. Governance: zero unresolved ethics, privacy or OPSEC incidents; review turnaround at or under 48 hours for 90% of submissions.

90-day plan

Resourcing. Crawl is deliberately low-cost: DCPA and Innovation and Emerging Technology time, one analytics lead at part time, ethics counsel on call, and a sensing-layer capability that can be an interim one (existing DCPA monitoring tools, platform-native analytics and hand-coded samples) for 90 days. The vendor question is a walk-phase question, but requirements and acquisition should be engaged during crawl so a contract vehicle exists when Gate 1 is passed.

What leadership is asked to approve

1. The proof of concept as bounded above, with the target chosen by the working group.

2. The governance standard in Section 7 as the program's rules, pending a DHA-specific instruction.

3. Named roles in Section 9, including the product owner's obligation to receive and disposition listening flags.

4. Early engagement of requirements and acquisition on a sensing-layer capability.

5. The executive sponsor's authority to seek written Department-level alignment on message vetting before the walk phase.

Risks and mitigations

The risks are real and mostly known; each has a control already in the standard or the playbook. The one risk that cannot be fully mitigated is that an authentic voice will say something true and unflattering. The program's answer is to fix the thing, not to silence the voice.

Open questions and decisions needed

Five decisions belong to the working group before the next leadership brief; four belong to leadership at the brief; three are items the group asked for and this paper partly answers.

For the working group

• Choose the proof-of-concept target from the three candidates in Section 11, with the product owner in the room

• Confirm whether a DHA-specific social media instruction exists; if not, decide whether writing one is a crawl deliverable

• Settle the question of Department-level alignment: who vets, what the ceiling is, and what written top cover looks like

• Decide the interim sensing-layer method for crawl (existing DCPA monitoring, platform analytics, hand-coded samples) and name the analytics lead

• Agree the dashboard layout (the deck's main or backup version) and the metric definitions with PI/P2P

For leadership

• Approve the bounded proof of concept and its success criteria

• Approve the governance standard as the program's rules

• Direct product owners to receive and disposition listening flags

• Authorize early engagement of requirements and acquisition on a sensing-layer capability

Items the group asked for

• Comparable service-industry and healthcare-adjacent examples: Section 5 supplies verified cases (Southwest, Marriott, Hilton, Mount Sinai, NYC Health + Hospitals, Cleveland Clinic, Dell, HHS, the Army, the Coast Guard, NASA). The honest finding stands: no healthcare organization has combined an employee-creator program with operational social listening at scale. Mount Sinai and Southwest are the two halves.

• Deck rework: the opening should state purpose, leadership relevance and expected outcomes before the information-environment slide; Section 1 of this paper is written to be that slide.

• The creator-tracking report: Section 4 and the companion Military Creator Pearls page. Two corrections to what was said on the call: the set is 30 creators across four platforms, not 30 platforms, and the report does not state a median follower count; its median figure is 1.02% Instagram engagement.

Acquisition and vendor considerations

• A sensing-layer vendor is a walk-phase purchase; requirements should describe outcomes (named-entity monitoring, flag routing, exportable samples, AI governance under DoDI 5400.19) rather than a product.

• Paid external creators require a contract vehicle; the UK GCS model (agency-run due diligence, mandatory disclosure, spend published in transparency reports) is the reference design.

• DoDI 5400.17 Section 7.1 permits third-party social media management platforms subject to acquisition and records rules; any tool must meet records-management requirements for official-account content.

• A LinkedIn-hosted live event, which DCPA has arranged with other agencies, costs nothing and can run in crawl.

References

Every link was opened and checked on 15 Oct 2026. Case-study figures are as reported by the organization or its named evaluator; self-reported figures are marked as such in Section 5.

Internal sources

• Top 30 U.S. Military and Veteran Creators: Cross-Platform Map (counts as of 15 Oct 2026) and Military_Influencer_Map_2026.xlsx

• DHA Digital Strategy and Influencer Plan working-group meeting, 15 Oct 2026 (recording summary and transcript)

• DCPAD working deck, Digital Workforce and Creator Strategy (October 2026 revision)

Policy and law

• DoDI 5400.17, Official Use of Social Media for Public Affairs Purposes, 12 Aug 2022, Change 2, 14 Feb 2025

• DoDI 5400.19, Public Affairs Use of Artificial Intelligence, 28 Jul 2025

• Joint Ethics Regulation, 15 May 2024

• SOCO Advisory 23-03, Application of Standards of Conduct to Personal Social Media Accounts, 5 May 2023

• 5 CFR Part 2635, Standards of Ethical Conduct for Employees of the Executive Branch

• 5 CFR Part 3601, Supplemental Standards of Ethical Conduct for Employees of the Department of Defense

• U.S. Office of Special Counsel, Hatch Act FAQs and Social Media Use Refresher, June 2021

• DoDD 1344.10, Political Activities by Members of the Armed Forces, 19 Feb 2008

• 16 CFR Part 255, Guides Concerning the Use of Endorsements and Testimonials in Advertising, 26 Jul 2023 and FTC, Disclosures 101 for Social Media Influencers

• 45 CFR 164.508, 45 CFR 164.502 and HHS guidance on marketing under the HIPAA Privacy Rule

• AR 360-1, The Army Public Affairs Program, 8 Oct 2020; ALNAV 024/25, Social Media Guidance for Uniformed Service Members of the Department of the Navy, 14 Mar 2025; Army social media policies

• Social Media in the Military Health System, health.mil; health.mil policy library search

• UK Government Communication Service, Guidelines for Influencer Marketing, Nov 2023

Case studies

• NASA Socials; JPL, anniversary of the first NASA Tweetup; Artemis II NASA Social

• SIGNAL, Army kicks off Creative Reserve pilot program; Army.mil, Operation Ring the Bell; Military.com, Army looks to tap soldier influencers; Business Insider, Army pilot paused

• Sherwood News, military deploying influencers for recruitment

• U.S. Coast Guard, Smarter Every Day highlights Coast Guard operations

• We Are The Mighty, Sam Eckholm; U.S. Veterans Magazine, Sam Eckholm

• HHS, We Can Do This: Live; COVID-19 Community Corps launch

• NACHC, Spring micro-influencer pilot report, 2022; AIM, Philadelphia engaging local influencers, 2025

• Shorty Awards, Mount Sinai Social Media Ambassador Program

• Hootsuite, NYC Health + Hospitals awarded Gold

• Content Marketing Institute, Cleveland Clinic content strategy

• Mayo Clinic News Network, Social Media Health Network; e-Patient Dave, Mayo social media network will sunset

• Southwest Airlines, Listening Center; PR News, Southwest social media on Bloody Sunday

• Shorty Awards, Marriott Bonvoy portfolio TikTok creators; Marketing Dive, Marriott TikTok correspondents

• Comscore, Hilton 10-minute TikTok

• Sephora, 2023 Sephora Squad

• The Drum, Duolingo's TikTok strategy

• Dell, social listening fact sheet; Dell, five years of SMaC University

• Social Media Today, LinkedIn Elevate employee-shared content

• 2025 Edelman Trust Barometer, Global Report; 2024 Edelman Trust Barometer Special Report, Trust at Work

Measurement

• AMEC Integrated Evaluation Framework; Barcelona Principles 4.0 summary, Open University

Creator-map sources (Social Blade and press pages for each figure in Section 4) are listed on the companion Military Creator Pearls page and in the spreadsheet.