1. How admissions worked before the automation
The team sold a changing timetable. Around 10 to 12 Course Advisers worked across New Zealand, Australia, Singapore and the United States. The provider offered Cyber Security, Data Science & AI and Software Engineering programmes in full-time and part-time formats, both remotely and in person. Cohorts rotated quarterly by region and university partner, with between 4 and 18 students in a class. Dates, prices, application cutoffs and upcoming webinars therefore had to be checked against the applicant's circumstances.
HubSpot did several jobs at once. Prospects entered as contacts and moved through values such as open, second call, third call, contacted, warm, hot, cold and qualified in one general lead-status field. That field mixed call attempts, level of interest and qualification. Ownership might come from automatic assignment or follow the adviser who booked and held the consultation. Once somebody enrolled, they left the active lead pipeline and moved into a separate student or customer process.
Most of the volume came before enrolment. Each adviser handled roughly 20 to 40 email enquiries a week, including contacts assigned from the shared inbox, while the Admissions Manager handled about 20 more directly. Across the team, there were about 40 to 80 calls and 1 to 6 booked hour-long consultations each day, leading to 3 to 12 enrolments a month. That left hundreds of contacts requiring follow-up, often across several channels.
- 20 to 40weekly enquiries per adviser
- 40 to 80team calls each day
- 1 to 6consultations each day
- 3 to 12monthly enrolments
2. Connecting the existing systems around the applicant
Every workflow began by identifying the person and the adviser responsible for them. n8n matched the email address or call to a HubSpot contact, read the relevant contact fields and recent activity, then combined that history with current course information from the shared sheets. What happened next depended on a real event the team already recognised: a new enquiry, a missed call, webinar attendance, a completed consultation or a confirmed enrolment.
The nine workflows behind this map are rebuilt on synthetic data as Exhibit 20, where each box is traced to the node that does it. They import into n8n and run without credentials.
From enquiry to Student Services
How a new enquiry, missed call, webinar attendance or completed consultation moves from HubSpot context to Student Services handover.
- automation
- human check
flowchart TD
accTitle: Admissions journey from first contact to Student Services
accDescr: Events gather CRM and course context, advisers review communication and enrolment, Admissions checks preparatory course completion for the alternative route, and staff confirm documents and payment terms before handover.
subgraph ENQ["1 · Enquiry and context"]
A["New email, call or webinar event"] --> B["Find the HubSpot contact and owning adviser"]
B --> C["Combine contact history with current course dates, cutoffs and webinars"]
C --> D["Prepare the next communication or task"]
D --> E{{"Adviser checks it"}}
end
subgraph FOLLOW["2 · Follow-up and admission"]
E --> F{"What happened?"}
F -->|Missed call or webinar| G["Place the contact in the matching follow-up sequence"]
F -->|Consultation complete| H["Prepare notes, email and next tasks"]
F -->|Alternative entry route| I["Track preparatory course completion"]
G --> J["Managed follow-up queue"]
H --> J
I --> K{{"Admissions checks completion evidence"}}
J --> L{{"Adviser confirms enrolment"}}
K --> L
end
subgraph HAND["3 · Enrolment handover"]
L --> M["Prepare agreement, payment details and student record"]
M --> N{{"Person checks documents and payment terms"}}
N --> O["Student Services receives the handover"]
end
classDef human fill:#f6ad55,stroke:#c05621,color:#1a1a1a;
class E,K,L,N human;Swipe to follow the workflow →
3. Drafting replies with the current course offer attached
The inbox watcher covered both the shared enquiry address and each adviser's account. For a new message, it found the matching HubSpot record and contact owner, read the useful history and properties, then retrieved the relevant course information. Dates, cutoffs and webinar schedules came from the sheets the team already maintained, rather than from a model's memory.
The writing guidance came from the sales team itself. I ran a structured analysis over thousands of earlier email conversations to identify how advisers answered recurring questions, when a consultation was more useful than sending a full course outline, and how the team explained price and payment plans. That became the playbook used to prepare a draft. The draft was added to the HubSpot contact record with the owner tagged, ready for the adviser to check and send.
Several follow-up journeys did not need AI at all. A missed first call started a recovery sequence. Webinar attendees and no-shows received different follow-up. A completed consultation created the agreed next tasks. Applicants without the standard degree prerequisite were directed to designated free preparatory courses; completion evidence then fed the organisation's routine approval process.
- Missed first call
Start the recovery sequence and create the next call task.
- Webinar attended
Send material that assumes the prospect saw the session and schedule adviser follow-up.
- Webinar missed
Send the alternative sequence rather than treating the registration as attendance.
- Consultation completed
Create the next task sequence using the outcome already recorded by the adviser.
4. Using one transcript more than once
Dialpad transcripts fed a single extraction workflow. The workflow prepared the HubSpot call note and pulled out what had been agreed: course, start date, payment plan, education level, delivery method and university partner. Those details filled the contact's HubSpot properties and, after a consultation, the team's follow-up email template with the right course outline, timetable and fee information. The adviser corrected transcript errors before anything became part of the customer record.

The same call could also support coaching. The analysis marked whether the adviser covered discovery, course detail, a tailored pitch, objections and the close. It compared how much the adviser and the prospect each spoke, grouped objections, and tracked phrases or topics that managers wanted to focus on that month. Those fields could then be compared across calls in charts and trend views without asking an adviser to recreate the conversation in a second system.
- Fixed workflow rules
Contact matching, assignment, reminders, scheduling, sequences and document movement.
- Language models
Email drafts, transcript extraction, CRM note drafts and coaching categories.
- Staff checks
Communication, transcript corrections, preparatory course completion, documents and payment terms.
5. Turning an open pipeline into a fixed daily queue
Managers needed a stricter mode during a performance review. When it was active, the workflow read the adviser's open, second-call and third-call contacts, applied timing rules and manager priorities, then created HubSpot tasks and fixed Google Calendar blocks. It deliberately removed the adviser's control over when those calls were attempted. Completed work updated the contact; unfinished work returned to the queue.
Manager-directed follow-up loop
How a manager-activated scheduling mode turns open, second-call and third-call contacts into a fixed daily queue.
- manager action
flowchart TD
accTitle: Manager-directed follow-up scheduling loop
accDescr: A manager activates structured scheduling, rules create a fixed adviser queue, and completion or non-completion sends each lead back through the next-action loop.
M{{"Manager activates the scheduling mode"}} --> A["Read open, second-call and third-call contacts"]
A --> B["Apply due-date rules and manager priorities"]
B --> C["Create HubSpot tasks and fixed Google Calendar blocks"]
C --> D["Adviser works through the prescribed queue"]
D --> E{"Was the task completed?"}
E -->|Yes| F["Update the lead status and next due action"]
E -->|No| G["Return unfinished work to the queue"]
F --> A
G --> A
classDef human fill:#f6ad55,stroke:#c05621,color:#1a1a1a;
class M human;Swipe to follow the workflow →
6. Carrying the sale into the student record
Enrolment moved the applicant into a different operating process. The active HubSpot lead became a student or customer record, and Student Services needed the decisions already made during the consultation. The workflow prepared contracts or student agreements, DocuSign documents, Google records, payment-plan fields and the relevant finance and calendar steps. It also drafted the adviser-to-Student Services handover from the full HubSpot history.
A person checked the final pack. The alternative-entry route was procedural rather than academic: staff confirmed that applicants had completed the designated preparatory study, then checked their documents and payment terms. The automation assembled the information but did not make those checks. That handover moved from roughly two working days to the same day.

7. What changed, and what can be checked
- Prospects waited less for a reply
Enquiry turnaround fell from just over 24 hours to about 10 hours; the figures are approximate.
- Student Services received the record sooner
Handover moved from two working days to the same day.
- Administrative work fell
One Student Services role was removed and another was redeployed.
- The existing tools remained in place
n8n connected the operation without asking the team to adopt another customer database.
The pattern is not specific to education. Service businesses often lose context at the same points: enquiry, appointment, follow-up, paperwork, payment and delivery handover. The useful question is which steps can follow fixed rules, which require interpreting language, and which still need a named person to decide.