Healthcare, family and paediatric chiropractic practice
GoHighLevel Patient Journey Automation: 13 Case Types, 6 Months
A designed and documented patient journey. The source records the workflow architecture; it does not record a verified deployment, patient outcomes or clinical results, and none are claimed here.
- Our role
- GoHighLevel automation architect
- Timeline
- Design and build engagement
- Sector
- Healthcare, family and paediatric chiropractic practice
At a glance
- Case types routed
- 13
- Automated touchpoints
- 20+ texts, emails and videos
- Journey length
- Six months, five phases
- Care phases unlocked by re-exam
- 3
- Platform
- GoHighLevel only (plus Zoom for virtual consults)
- Workflows per stage
- One, with branches, not 13 copies
What was breaking
Chiropractic care for families is not one service. An infant, a pregnant mother, a child with neurodevelopmental needs, a teenager with anxiety and an adult after a collision are five different clinical conversations with different language, different reassurance, different timelines and, frankly, different fears. This practice served thirteen recognisably distinct case types and had one message sequence for all of them.
The generic messaging was the visible problem. The structural one was that staff moved patients between stages by hand. That matters more than it sounds: if a stage change is a manual drag, then a reminder specified as 'twenty-four hours before the examination' has nothing to count from. There is no appointment time in the system for the automation to reference, so the most useful reminders in the entire journey simply cannot be built reliably.
There was also a scale trap waiting. The obvious way to personalise thirteen case types is to build thirteen journeys. For a six-month journey with five phases and twenty-plus touchpoints. That is somewhere north of two hundred and sixty workflows to build, test and then maintain forever, where every future copy change means finding and editing thirteen places. The practice would have ended up with a system nobody could safely modify.
Identical messaging to clinically different patients
A third-trimester mother and an adult after a collision received the same texts. Generic communication in healthcare does not read as neutral; it reads as inattentive.
Reminders had nothing to count from
Because stage changes were manual drags rather than booked appointments, time-relative reminders like '24 hours before' had no anchor date available to the automation.
Staff time spent moving patients between stages
Administrative dragging is both a cost and a single point of failure: a forgotten drag silently stops a patient's whole journey.
Thirteen-fold duplication was the default answer
Personalising by duplicating workflows would have produced hundreds of near-identical workflows, making every future copy change a thirteen-place edit with one inevitably missed.
Care-plan adherence depends on contact between visits
The gap between appointments is where patients disengage. With no automated check-ins, adherence rested on whatever the patient remembered from the last visit.
What we built
The whole design turns on one decision: capture the case type once, convert it to a tag, and branch on that tag inside a single workflow per stage. The booking form gains a case-type dropdown. A small intake workflow converts that field into a tag. Every stage workflow then opens with a conditional branch reading those tags. Thirteen case types cost thirteen branches inside one workflow instead of thirteen workflows, and a copy change happens in one place.
Critically, every branch set includes a default. Any patient who arrives untagged, or whose case type has not had its copy written yet, falls through to a sensible general message rather than silence. This is what made the build shippable before all thirteen sets of copy existed: the practice could go live with the default covering everyone and fill in specific branches one at a time, with each addition improving the system rather than being a prerequisite for it.
I fixed the reminder-anchor problem by changing what the automation counts from. Instead of trying to schedule relative to a manual stage drag, re-examination reminders key off a booked calendar appointment. Adding a step that records the re-exam as scheduled gives every downstream reminder a real date to reference. The same approach solved the Zoom problem: the meeting link is sent by a calendar reminder conditioned on the appointment type, so virtual consultations get a link and in-person ones do not.
The journey itself runs in five phases. The new-patient phase covers the consultation and examination (welcome and paperwork on booking, a what-to-expect video the day before, the Zoom link fifteen minutes ahead for virtual consults, a personal note from the practitioner after the consultation, an office-tour video before the examination, and a recap after the report of findings. Start of care runs on waits from sign-up: first adjustment, a message after the second visit acknowledging that healing is not linear, then two-week and four-week check-ins. Phases one and two each hinge on a re-examination) preparation video on booking, reminder twenty-four hours out, prescribed exercises afterwards, then check-ins at one and three weeks. Phase three carries the patient to the six-month progress consultation.
On top of the case-type routing sits a second personalisation layer: whether the patient is an adult or a child, whether exercises were assigned, and whether a child is neurotypical. These are separate tags checked by the same conditional branches, which is why a family with children receives an additional email during phase one while an adult on the same clinical pathway does not. Layering independent tags rather than multiplying case types is what keeps the combinatorics from exploding.
One smaller fix is worth naming because it shows how design and data model interact. The practitioner wanted to send a personal note after a consultation, and there was no stage representing 'consultation completed' for that message to hang on. The answer was not a workaround in the automation. It was adding the missing stage. When an automation is hard to express, the pipeline is usually wrong.
One workflow per stage, branches inside
Thirteen case types become thirteen conditional branches in a single workflow rather than thirteen duplicated workflows. The alternative is a system that works on day one and cannot be safely changed on day ninety.
Always include a default branch
Every branch set catches untagged patients with a general message. This is what let the practice launch before all thirteen sets of copy were written; the default is a safety net, not a placeholder.
Count from booked appointments, not manual drags
Time-relative reminders need a real date. Anchoring them to a booked calendar appointment rather than a staff action is the difference between reminders that work and reminders that silently never send.
Layer independent tags instead of multiplying case types
Adult-or-child, exercises-assigned and neurotypical are separate tags checked by the same branches. Folding them into the case-type list would have turned thirteen options into dozens.
If the automation is awkward, fix the pipeline
A message with no stage to attach to meant a stage was missing, not that the automation needed a workaround. Adding the stage was simpler and left the data model more honest.
Non-negotiables
- GoHighLevel only. The practice did not want a second system to pay for or learn; the only non-native element is Zoom for virtual consultations.
- Clinical judgement stays with the practitioner. Automation handles timing, education and reminders, never assessment or advice.
- Copy for thirteen case types could not all be written at once, so the design had to work while branches were still being filled in.
- The practice's existing booking flow had to keep working; case type had to be captured without adding friction to booking.
The pipeline
Drawn as text and shapes rather than shipped as an image, so the step labels stay readable and downloadable.
The six-month patient journey
Five phases, each a single workflow whose opening branch reads the patient's tags. Re-exam reminders count from booked appointments, which is what makes them reliable.
- Trigger
- Processing step
- Branch / filter
- Error handling
- Data store or CRM
- Delivered output
- Primary path
- Conditional or exception path
The personalisation layer
Independent tags applied at intake and checked by the same branches inside every stage workflow, which is why the case-type count does not multiply.
- Trigger
- Branch / filter
- Processing step
- Delivered output
- Primary path
How the work runs
Each stage, what it does, and the specific failure it is built to prevent.
Case type is captured once, at booking
The booking form gains a case-type dropdown. This is the only place the information is collected, and it is collected before the patient has met anyone, which means every subsequent message can be correct from the first contact rather than from the first appointment.
A small intake workflow converts the field into a tag
Form fields are awkward to branch on across many workflows; tags are not. One short workflow translates the captured case type into a tag, and from then on every stage workflow reads tags. Doing this conversion once, centrally, means the thirteen-way logic exists in one readable place.
Keeps branching logic uniform, every workflow asks the same kind of question in the same way.
Every stage workflow opens with a branch and a default
Each of the five phases is one workflow that begins by checking the patient's tags. The default branch is the important part: a patient with no tag, or whose case type has no copy yet, receives sensible general messaging. The practice could launch immediately and write specific branches over time.
No patient can fall into silence because a branch has not been written yet.
Re-examinations are recorded as booked appointments
This is the fix for the reminder-anchor problem. Rather than scheduling relative to a manual stage change, the design records the re-exam as a booked calendar appointment, giving downstream reminders a real date. Without this step, 'twenty-four hours before the re-exam' is unbuildable.
Removes the dependency on a staff member remembering to drag a card at the right moment.
Virtual consultations get a conditional meeting link
A Zoom link is only meaningful for a virtual appointment, so it is sent by a calendar reminder conditioned on appointment type, fifteen minutes before the consultation. In-person patients do not receive a link they would find confusing.
Start of care runs on waits, not stages
The first weeks of care are a fixed cadence from sign-up rather than a sequence of stage changes, so they run on timed waits: the first adjustment, a message after the second visit acknowledging that healing is not linear, then two-week and four-week check-ins. That particular message exists because the second visit is where patients commonly lose confidence.
Each care phase hinges on a re-examination
Phases one and two follow the same shape: a preparation video when the re-exam is booked, a reminder a day before, prescribed exercises afterwards, and check-ins at one and three weeks. Exercise check-ins only fire when an exercises tag is present, so patients who were not given exercises are not asked about them.
Tag-gated sequences prevent the single most irritating automation failure: being asked about something you were never given.
Family-specific messaging layers on top
Tags for adult-or-child and for neurotypical children add an extra phase-one email for families, inside the same workflows. These are independent layers rather than new case types, which is the design decision that stops thirteen options from becoming dozens.
A missing stage gets added rather than worked around
The practitioner wanted to send a personal note after the consultation, and no stage represented that moment. Rather than contorting a workflow, the design adds a consultation-completed stage. When a message has nowhere natural to attach, the pipeline is usually incomplete.
What changed
| Aspect | Before | After |
|---|---|---|
| Message relevance | Every patient received the same texts regardless of case type or age. | Thirteen case types, plus adult/child and family layers, each receive copy written for them. |
| Reminder reliability | Time-relative reminders had no appointment date to count from, so the most useful ones could not be built. | Re-exams are recorded as booked appointments, giving every reminder a real anchor date. |
| Stage movement | Staff moved patients between stages by hand; a forgotten drag silently halted a journey. | Calendar and stage triggers drive the journey, with manual drags no longer load-bearing for timing. |
| Maintainability | Personalising meant duplicating workflows; a copy change would have been a thirteen-place edit. | One workflow per stage with branches inside; a copy change happens once. |
| Contact between visits | The gap between appointments was unmanaged, leaving adherence to memory. | Timed check-ins at two weeks, four weeks, and one and three weeks after each re-exam. |
| Launch risk | Nothing could go live until copy existed for every case type. | A default branch covers everyone, so branches are added incrementally after launch. |
What it runs on
Every tool listed with the job it actually does in this build.
Platform
GoHighLevel
The entire build, pipelines and stages, workflow triggers, calendars and booking forms, custom fields, tags, conditional branches, waits, SMS and email in a single workflow, and hidden site pages hosting the patient videos.
Orchestration
GoHighLevel Calendars
The anchor for every time-relative reminder. Booked appointments are what the 24-hour reminders and meeting links count from.
Integration
Tags and conditional branches
The personalisation mechanism: case type, adult or child, exercises assigned and neurotypical status, all checked at the top of each stage workflow.
Delivery
Zoom
The only non-native element, for virtual consultations. The meeting link is injected by a calendar reminder conditioned on appointment type.
What stopped being manual
Snippets are illustrative and generic. No client data appears in any of them.
Thirteen case types in one workflow
Replaces the alternative of building and maintaining thirteen parallel journeys. The manual act removed is not just sending messages. It is the ongoing work of making the same copy change in thirteen places and discovering months later that one was missed.
text
// ONE workflow per stage. Thirteen branches, not thirteen workflows.
stage: "Report of Findings"
if tag = case-pregnancy -> pregnancy recap copy
if tag = case-paediatric -> parent-facing recap copy
if tag = case-accident -> recovery-focused recap copy
... (13 branches)
else -> DEFAULT general recap copy
// covers untagged patients and
// branches not yet writtenGiving reminders something to count from
Replaces a staff member dragging a card and hoping the timing works out. A reminder specified as '24 hours before' needs an actual date; recording the re-exam as a booked appointment supplies one.
text
// BEFORE: reminder relative to a manual stage drag
// -> no appointment time exists -> reminder cannot be scheduled
// AFTER: re-exam recorded as a booked calendar appointment
on appointment.booked -> send preparation video
24h before appointment -> send reminder // real anchor date
on appointment.completed -> send exercises [if tag: exercises]Conditional meeting links
Replaces someone checking whether an appointment is virtual before sending a link, and avoids the opposite failure of sending a Zoom link to a patient who is driving to the clinic.
text
15 min before appointment:
if appointment.type = virtual
-> send {{appointment.meeting_link}}
else
-> send parking and arrival detailsWhat it was worth
The design's value is mostly in what it avoids. The obvious build, one journey per case type, would have produced a system that worked on launch day and became unmaintainable within months, because every copy change would have been a thirteen-place edit with no way to verify you had found them all. Branching inside a single workflow per stage keeps the whole six-month journey in five editable places.
The second real gain is that reminders became possible at all. Anchoring them to booked appointments rather than manual stage changes turned the most clinically useful messages in the journey (preparation before a re-examination, a reminder the day before) from something that could not be built reliably into something that runs on its own.
And the default branch is what made the project shippable. Without it, launch depends on thirteen sets of finished copy; with it, the practice goes live covering every patient and improves the system incrementally. That is a sequencing decision rather than a technical one, and it is usually the difference between a design that ships and a design that waits.
I want to be straightforward about the limits of this entry: it is a design and build case study. The source documents the architecture, the case types and the touchpoint count. It does not document patient outcomes, retention, or adherence figures, so none appear here. A results table full of invented numbers would be worth less than an honest architecture.
Case types routed
Each a conditional branch, not a duplicated workflow.
Automated touchpoints
Texts, emails and videos across the six-month journey.
Journey phases
New patient, start of care, and three care phases.
Care phases unlocked by re-examination
Each with preparation, reminder, exercises and check-ins.
Workflow per stage
Instead of one per stage per case type, the decision that keeps it maintainable.
Independent personalisation tags
Layered rather than multiplied into the case-type list.
Handed over
- A six-month patient journey across five phases, from consultation booking through to the final progress consultation.
- Thirteen case types routed through conditional branches inside one workflow per stage, each with a default branch.
- More than twenty automated touchpoints spanning SMS, email and video.
- A personalisation layer of independent tags: case type, adult or child, exercises assigned, and neurotypical child.
- Re-examination scheduling recorded as booked calendar appointments, so time-relative reminders have a reliable anchor.
- Conditional Zoom meeting links for virtual consultations only, sent fifteen minutes before the appointment.
- A consultation-completed pipeline stage added so the practitioner's personal note has a natural trigger.
- Exercise check-in sequences gated on an exercises tag, so patients are only asked about exercises they were given.
Questions about this build
What teams usually ask when they recognise their own process here.
Because it does not survive contact with maintenance. Thirteen case types across five phases and twenty-plus touchpoints is well over two hundred workflows, and every future copy change becomes a thirteen-place edit where a missed one is invisible. Branching inside one workflow per stage gives identical patient-facing behaviour while keeping the whole journey in five editable places. The duplicated version is faster to build and far more expensive to own.
By layering independent tags instead of enumerating combinations. Case type, adult-or-child, exercises-assigned and neurotypical are four separate questions checked by the same conditional branches. If you fold them into the case-type list you get a combinatorial explosion; keep them independent and you get four small checks that compose. Adding a fifth layer later costs one check, not thirteen new branches.
A reminder like '24 hours before the re-examination' needs a date to count backwards from. When stage changes are manual drags, no appointment time exists in the system, so the reminder has no anchor and cannot be scheduled reliably. The fix is to record the re-exam as a booked calendar appointment, which gives every downstream reminder a real date. It is a small change to the data model that unlocks the most useful messages in the journey.
Two things. It catches patients who arrive untagged, so nobody falls into silence because of a data gap. And it makes the build shippable before all thirteen sets of copy are written, the practice launches with the default covering everyone and adds specific branches over time. The caveat is that an unwritten branch looks like a working system, so it is worth tracking how many patients land in the default.
For scheduling, education and check-ins, yes, provided the boundary is explicit. Everything in this journey is timing, preparation material, reminders and adherence contact. Nothing assesses a patient or offers clinical advice, and nothing replaces a conversation with the practitioner. The value is that the gap between appointments stops being unmanaged; the line that must stay firm is that automation never implies clinical judgement.
I can only answer the first half honestly: this is a designed and documented solution, and the source material does not record a verified deployment or any patient outcome data. So I am not going to attach retention or adherence figures to it. What the entry demonstrates is the architecture, the branching strategy, the reminder anchoring, and the sequencing decision that let it launch incrementally.
Capabilities this used
Read next
All gohighlevel services workHave a process that works like this one did?
Book a strategy call. We will walk the workflow with you and say plainly which parts are worth automating and which are not.