A network is as strong as its weakest month
A government program delivered through forty partner sites is not one program. It is forty organizations with different boards, different payroll cycles, different software and different ideas of what the contract says. The government customer sees one set of results and holds one party accountable for them. The job of program management is to make forty organizations behave enough like one that the results are real, without crushing the local knowledge that made each partner worth funding in the first place.
That job is mostly structure: who each site calls, what each site follows, where everything lives, how performance is judged, what happens when a site slips, and how fast the money moves. Get the structure right and the program manager spends the week on the three sites that need attention. Get it wrong and the week goes to the inbox.
Organize around regional leads, not a central inbox
A central team cannot hold a working relationship with forty sites. It can process their paperwork. The relationship belongs to a regional lead: one named person each site calls first, who knows the site director by name, has visited in person and can tell the difference between a site having a bad month and a site in trouble.
As a working range, one regional lead can carry eight to twelve sites well, fewer if the sites are new or spread across long drive times. Past that range, the lead stops visiting and starts triaging, and early warning disappears. The central team keeps the functions that need consistency: the partner agreements, payments, data and reporting, compliance monitoring and the relationship with the government customer.
Write the split down. For each recurring activity, name who does it, who approves it and who needs to know. Partners should be able to answer "who do I call about this" without asking.
Standard operating procedures sites actually follow
A forty-page operations manual is read once, at onboarding, by the person who will leave in six months. Procedures that sites follow are short, specific, attached to the form they govern and written for the person doing the task. Each one fits on two pages, shows a completed example and states what good looks like.
- Intake and eligibility. What to verify, which documents count, what to do when a document is missing, and who can approve an exception.
- Service documentation. What a complete service record contains, the deadline for entering it and the data definitions that apply.
- Incident reporting. What counts as an incident, who is called, within how many hours, and what is written down afterward.
- Invoicing. What to submit, by when, with which backup, and how long payment takes once an invoice is accepted.
- Staff onboarding. Required training, background checks where the program needs them, system access requests and the first-week checklist.
- Records and privacy. What may be stored where, who may see it, how long it is kept and how it is destroyed.
Version every procedure, date every change and announce changes on the monthly call with a one-line summary of what is different. When a procedure changes, the old version stays in the library marked as superseded, so a monitor reviewing last spring's files knows which rule applied then.
One portal, one source of truth
Email attachments are where program records go to disappear. The second version of a form, the corrected invoice, the signed agreement amendment: each lives in one person's inbox and nowhere else. A distributed network needs a single partner portal, and it needs to be the only place things count.
The portal holds the current procedures and forms, the partner agreement and amendments, monthly data submission, invoice submission with status that partners can see, an issue log, announcements and the partner's own performance summary. One login per person, with roles: site director, site staff, finance contact, regional lead, central team. Nothing complicated. A partner should be able to see in one screen what is due, what is late, what has been paid and what is waiting on whom.
Then hold the line. When a partner emails an invoice, the answer is a friendly link to the portal, every time. Within two cycles, everyone uses it.
The monthly partner call
One call a month with every site, at the same time, for no more than an hour. It is the only moment the whole network hears the same thing at once, so it carries the announcements, the procedure changes and the deadlines. It also carries something more useful: one site, each month, spending ten minutes on how it solved a problem the others have.
A workable agenda: network results for the month, five minutes; changes and deadlines, ten; the peer share, ten; open questions, the rest. Record it, post the notes and the recording to the portal within a day and track attendance. Sites that miss two calls in a row get a call from their regional lead, not a warning.
Do not use the group call for performance problems. Those belong in a private conversation between the site and its regional lead, the same week the problem shows up in the data.
Performance tiers and what each one triggers
Partners deserve to know how they are judged and what happens next, before anything happens. Publish a short scorecard: timely reporting, data quality, service volume against target, compliance findings from monitoring and the program's primary outcome. Assign a tier on a trailing three-month window, so one bad month does not move a site and three bad months do.
| Tier | What it looks like | Response | Contact rhythm |
|---|---|---|---|
| Leading | On time, clean data, at or above target, no open findings | Ask them to lead a peer share, offer them first call on expansion and pilot new procedures with them | Quarterly check-in |
| Meeting | Meets the scorecard with minor, self-corrected issues | Routine support and the standard monitoring schedule | Monthly regional lead contact |
| Watch | One measure below standard for the window, or a new compliance finding | Regional lead conversation within a week, a written note of the issue and an agreed fix, technical assistance offered | Every two weeks until resolved |
| Corrective action | Two or more measures below standard, a repeat finding, or a Watch issue that did not close | Formal corrective action plan with milestones and evidence of closure | Every two weeks, documented |
| Transition | A corrective action plan failed, or a serious finding involving participant safety or misuse of funds | Escalation to the government customer, a plan to move participants and records, and closeout of the site | Weekly until closed |
The top tier matters as much as the bottom. Sites that perform well and hear nothing learn that performance is invisible. Recognition costs nothing, and offering strong sites first call on new funding is the clearest signal the network can send about what it values.
Corrective action plans that close
Most corrective action plans fail in one of two ways. They are vague, so nobody can say when they are done, or they are written by the network manager and signed by the partner, so the partner never owned the fix. A plan that closes is specific, measurable, written by the partner and checked against evidence.
A Corrective Action Plan in Six Steps
Set the windows in the partner agreement, not at the moment of the problem, so the process feels like a rule and not a reaction. Allow one extension when the partner is making real progress. A second extension is a decision not to decide, and every other site in the network notices it.
The cause analysis is the step that earns the plan its value. A site behind on data entry may have a staffing gap, a training gap, a bad internet connection or a director who does not believe the data matters. Each needs a different fix. A plan that skips the cause tends to prescribe "additional training" for all four.
Pay on time, every time
Community partners run on thin cash. A forty-five day delay on a reimbursement can mean a missed payroll, a line of credit drawn down or a director deciding the program is not worth renewing. Late payment is the most expensive thing a network manager can do, because the bill is paid by the best partners leaving first. The partners with options leave soonest, and they are usually the strong ones.
Treat payment speed as a performance measure for the network itself. Publish the clock: how many business days from an accepted invoice to payment. Track the actual number every month and report it to partners on the monthly call. When an invoice has a disputed line, pay the undisputed lines on schedule and resolve the rest separately. Never hold a full invoice over a small question.
Help partners get invoices accepted the first time. A short invoicing procedure with a completed example, a checklist of required backup and a pre-submission review for new partners removes most rejections. Where the program's terms permit advance payments or a working capital advance for small partners, use them for the first months, when a new site's cash flow is tightest.
Escalation that does not depend on who you know
In a network without a written escalation path, problems move at the speed of personal relationships. A site director who knows someone at headquarters gets an answer in a day; a new director waits three weeks. Write the path down, with response times, and apply it the same way to everyone.
A simple version has three levels. The regional lead responds within one business day and resolves most things. The program manager takes what the regional lead cannot resolve within a week, and anything involving money, a policy interpretation or a dispute. The government customer is brought in for anything touching scope, funding, a change to the program's terms, or a serious compliance or safety issue. Participant safety skips the levels entirely: the site calls the program manager directly, the same day.
Every issue goes in the portal's issue log with a date opened, a severity, an owner and a date closed. The log turns escalation from a feeling into a record, and it shows which problems recur across sites, which usually means the procedure is wrong rather than the sites.
Documentation the government customer can rely on
The government customer cannot visit forty sites. What it knows about the program is what the documentation tells it, so the documentation has to be complete, consistent and honest about problems.
Send a monthly program report that tracks the contract's deliverables line by line: the active site roster, each site's tier, open and closed corrective action plans, payment timeliness, open issues by severity, risks with their mitigation, and results against the targets in the contract. Report a problem the month it appears, with the plan, rather than the quarter it becomes undeniable. A customer that learns about a failing site from you trusts the next report; one that learns about it from a complaint does not.
Keep the files a monitor or an auditor would ask for: site visit reports, monitoring results, approvals of exceptions, procedure versions and the decisions behind any change to a site's status. Remember who can change what. Program staff coordinate with the government's technical representative day to day, but only the contracting officer changes the scope, the price or the terms. Any request that would change those goes to the contracting officer in writing before anyone acts on it.
Scaling the network up or down
Growth is where networks lose quality. Ten new sites onboarded in one month overwhelm the regional leads and the portal at once, and the first month's data from all ten is unusable. Onboard in cohorts of three to five, with a readiness checklist each site completes before its first participant: agreement signed, staff trained, system access working, a test record entered and a test invoice accepted. Pair each new site with a leading site for its first quarter.
Shrinking the network needs as much planning. When funding drops or a site exits, the participants come first: a transition plan that names where each active participant goes, how their records move under the privacy terms, and who tells them. Then the money and the paper: final invoices, return of equipment and closeout of the site's records. Choose which sites to keep on their tier history, documented in advance, not on proximity or relationships. The rest of the network is watching how the decision is made.
A checklist for the first ninety days
- Every site has a named regional lead who has visited in person
- A written split of who does, approves and is informed for each recurring activity
- Two-page procedures for intake, documentation, incidents, invoicing, onboarding and records, each with an example
- One portal for documents, data, invoices and issues, with role-based logins
- A monthly partner call on a fixed schedule, with notes posted within a day
- A published scorecard, tier definitions and corrective action windows in every partner agreement
- A published payment clock, tracked monthly and reported to partners
- A three-level escalation path with response times and a same-day path for participant safety
- A monthly program report mapped to the contract's deliverables
- A cohort onboarding checklist and a written transition plan for any site that exits
Frequently asked questions
As a working range, eight to twelve established sites, fewer if the sites are new, struggling or far apart. The limit is visits and early warning: past that range, the lead stops knowing each site well enough to spot trouble before it shows up in the data.
Each finding, its cause, the action that fixes it, a named owner, a due date and the evidence that will show it is closed. The partner writes it, the network agrees it and adds support, and closure is checked against records or data rather than taken on assurance.
Pay them quickly and predictably, recognize performance publicly, offer them first call on expansion and new funding, and keep reporting light enough that it does not consume their best staff. Late payment and invisible performance push out strong partners first, because they have other options.
In the monthly report for the month the problem appears, with the plan to fix it. Anything involving participant safety, misuse of funds or a change to scope goes to the customer immediately through the agreed escalation path, and scope or price changes go to the contracting officer in writing.