A practice with three sites and one main number gets a text from a patient who wants to move Thursday's visit. The reply lands in the queue for the site she used last. That site is down two front desk staff this week, so the text sits while the desk across town has an empty queue.
She answered within minutes. Where her reply went next decided whether she got an answer that morning or called the office at noon, adding a call to a desk already behind.
Patient messaging workflows are the rules behind that path. They set which inbox a reply enters, who owns it, how long they have, what writes back to the EHR, and what gets logged. Our position is that the routing model comes first. Every other rule depends on it, and it's the hardest one to change once staff build habits around it.
Single-site habits break first at the handoffs. A text gets answered by the wrong site, a reschedule lands at a location the patient never visits, or an opt-out gets honored at one desk and missed at another.
This guide covers the design layer for multi-location practices: routing models, ownership and escalation, EHR write-back, and governance for templates, consent, and audit. It stays with the operating model every message type runs on.
Expect leadership to ask about phone call volume within weeks of launch. The last section ends with five monthly checks an administrator can run without IT, and the first one shows which site is falling behind.
Specific workflows, like reminders and refills, are covered in patient texting workflows that cut front desk calls.
A patient messaging workflow has five parts, and each needs a written rule before launch.
Several locations change the design because a reply has to find a specific person, and one site's desk can't see another site's workload. A single clinic can still improvise.
With three sites, a reply could belong to any of three desks. In the MGMA Stat poll on phone bottlenecks in medical practices (294 responses, March 10, 2026), 31% of leaders named scheduling their biggest staff-time drain on the phones. Many messaging software challenges at multi-site practices start at this point.
A routing model is the rule that picks which inbox a reply enters.
Location-based routing sends a reply to the inbox of the site tied to the patient's appointment. Skill-based routing sends it to the staff group trained for that request, such as scheduling, billing, or referrals.
Location-based routing fits independent site teams that run their own schedules and hours. Skill-based routing fits a central team that books for every site, where the scheduler's training matters more than which building the patient visits. The central-team version is laid out in call center workflows.
One inbox per site gives each desk clear ownership and clean site-level reporting, but coverage fails when a site runs short. A single shared inbox covers every gap, since any trained staff member can answer. Nobody owns a thread by default, though, and site numbers blur.
In an illustrative per-site week, the south site's median reply time drifts from 10 to 45 minutes. In a shared inbox, that drift disappears into the group average. Patient messaging software handles either setup, so staffing decides.
Routing models compared
|
Model |
Best fit |
Main risk |
|---|---|---|
|
One inbox per location |
Independent site teams |
No coverage when a site is short |
|
Single shared inbox |
Central scheduling team |
Nobody owns a thread by default |
|
Skill-based routing |
Multi-specialty groups |
Rules drift as staff change |
|
Location plus overflow |
Most multi-site practices |
Needs a maintained threshold |
An overflow rule moves new replies from a site's inbox to a backup team once that site's queue passes a set threshold.
In an illustrative version, when a site has more than 15 unanswered threads, or any thread older than 30 minutes, new replies route to the central team. Routing returns to the site once its queue drops below 8. Two numbers stop threads from bouncing back and forth. Review both each quarter, since staffing changes make last quarter's numbers wrong.
A response target is the time a practice commits to answering a message.
An unanswered thread belongs to the person assigned when the thread opens, before anyone replies. Assignment on first reply leaves new threads without an owner until someone picks them up, so the hardest questions wait longest, often a billing dispute or refund.
Assignment on open puts a name on every thread the moment it arrives, based on the routing rule. That name is who the escalation timer checks. It's also who the audit shows if the thread sat for three hours.
An escalation timer holds when three pieces work together: the timer, a named second owner, and a queue both people can see. A timer on its own sends an alert that no one is assigned to act on.
Naming the second owner, such as the site's front desk lead, keeps the alert from going to everyone at once. The visible queue lets that lead spot aging threads before the timer fires. Illustrative settings: 30 minutes for scheduling threads, 2 hours for billing.
An after-hours handoff tells the patient when a person will read the message and where to go if it can't wait. A workable auto-reply reads: "[Practice name] is closed. We'll reply after 8 AM. For an emergency, call 911."
The auto-reply must not promise a clinical callback, confirm a medication change, or suggest a symptom can wait until morning. Overnight threads open in the morning queue with their original timestamps, so staff answer the oldest first. That same message can serve every site.
Write-back is the step where a patient's reply updates the EHR without anyone retyping it.
Three things must write back: appointment status, cancellation reason, and the message record. If status doesn't post, a patient who texted C still shows as unconfirmed, and staff calls anyway, adding phone call volume.
A missing cancellation reason hides whether a ride fell through or the patient switched practices. Skip the message record, and the chart can't show what the patient was told. Before launch, book, confirm, and cancel a test patient, then check what your patient workflow automation writes at each step.
Clinical content belongs in the chart, and logistics can stay in the message thread. The dividing test is one question: would a clinician need this to make a care decision later?
A patient reporting new chest pain passes and goes into the chart. "Can I come at 10 instead?" stays in the thread. HIPAA's designated record set covers medical and billing records plus other records used to make decisions about patients (45 CFR 164.501), so messages used that way belong with the record.
Governance is the set of rules that keeps workflows consistent after launch.
A template library by site is a set of central, approved messages with a few local fields, such as site address, parking note, and callback number. Free typing at the desk defeats the audit, because every staff member words the same message differently and nobody can check what patients were told.
Central templates also keep healthcare texting inside the FCC's 160-character limit for exempt messages, and consistent with the message types registered for 10DLC. Sites request new templates through the administrator.
Opt-out records must live in the messaging platform so a STOP sent to one location stops texts from all of them. FCC rules at 47 CFR 64.1200 cap exempt healthcare texts to cell phones at one a day and three a week, with STOP honored immediately.
For home landlines, the FCC's TRACED Act report and order (FCC 20-186), adopted December 29, 2020, set a matching cap on exempt healthcare robocalls. Since April 11, 2025, a TCPA consent revocation must be honored within 10 business days.
Five monthly checks cover patient messaging workflows, and an administrator can run all of them from platform reports and the EHR schedule.
Log each result in one sheet by site, so a slipping location shows up by the second month.
Curogram is a HIPAA-compliant, SOC 2 Type II patient communication platform. Custom Filter Groups let admins set preset teams, such as Call Center Team 1, and shift routing rules as staffing changes. Messages route by location or provider, and admins edit automated templates for each location.
Patient messaging workflows hold across locations when every reply has a path, an owner, and a clock.
Choose the routing model first and write the overflow rule with real numbers. Inbound call reduction follows, because fewer replies turn into callbacks.
Book a Curogram demo and bring a map of your sites and inboxes.
Frequently Asked Questions