Training & education · 2026
Lead capture and sales follow-up automation
An automation layer for a training centre that sells courses through social ads. Every lead is captured within seconds, de-duplicated, scored and assigned to an advisor, and follow-up messages go out on WhatsApp and email until the lead books a call or opts out.
Built by
- Client
- Withheld at the client’s request
- Lead sources
- Meta ads · Website · WhatsApp
- Follow-up
- WhatsApp · Email
- Typical timeline
- 6–8 weeks
- Squad
- 5 people · 4 disciplines
Technologies
- n8n
- Laravel
- Next.js
- MySQL
- Redis
- Meta Lead Ads API
- WhatsApp Cloud API
- LLM APIs
The product
10 images01
Brief
Leads arrived faster than anyone could call them.
Context
The centre runs ads for several courses at once. Leads came from forms, the website and WhatsApp into spreadsheets that advisors shared.
Problem
Some leads were called twice, many were never called, and nobody knew which ads brought students rather than clicks.
Goals
- Every lead captured in seconds
- One owner per lead
- Follow-ups that never forget
- Cost per enrolment by campaign
02
Team
The squad behind it.
| Tech leadLead flow design and the CRM data model | ×1 | Lead flow design and the CRM data model | Senior | |
|---|---|---|---|---|
| Product designerAdvisor pipeline and the manager dashboard | ×1 | Advisor pipeline and the manager dashboard | Senior | |
| Frontend engineerCRM screens and dashboards | ×1 | CRM screens and dashboards | Senior | |
| Backend & automation engineern8n workflows, integrations and lead scoring | ×1 | n8n workflows, integrations and lead scoring | Senior | |
| QA engineerWorkflow tests and message checks | ×1 | Workflow tests and message checks | Senior |
03
UI/UX design
An advisor’s day on one screen.
Who worked on it
Advisors open a list sorted by who to call now: new leads first, then follow-ups due today. Each lead shows its course, its source and a short summary of past messages, written by the assistant.
- Call-next listNo hunting through spreadsheets.
- One-line historyThe whole conversation summarised.
- Outcome in two tapsInterested, call later, enrolled or lost.
04
Frontend
Managers see the funnel, not the noise.
Who worked on it
A dashboard follows each campaign from lead to enrolment, with response times per advisor and leads waiting too long flagged in red.
- Campaign to enrolmentReal cost per student, not per click.
- Response timePer advisor and per course.
- Waiting leadsFlagged before they go cold.
05
Backend & architecture
Workflows in n8n, rules in code.
Who worked on it
n8n receives leads from ad forms, the website and WhatsApp, and calls the CRM’s API to create or merge them. The CRM assigns each lead by course and advisor load, a model scores intent from the first messages, and scheduled workflows send follow-ups until the lead replies.
Decision 1
The CRM owns the data; n8n only moves it
Assignment and duplicate rules must live in one place.
Trade-off · Changing a rule means a CRM release, not a workflow edit.
Decision 2
Duplicates merged by phone number
The same person fills several forms across campaigns.
Trade-off · Shared family numbers need a manual split.
Decision 3
Approved message templates only
WhatsApp rules and the brand’s tone are respected.
Trade-off · New wording needs approval before it is used.
06
QA
Every workflow tested with test leads.
Who worked on it
Test leads go through every path — new, duplicate, opted out, no reply — in a staging copy of the CRM and a WhatsApp sandbox, before any workflow change reaches real people.
- Path coverageEvery branch of every workflow.
- Opt-out respectedTested on every release.
- Message previewArabic and English templates checked.
Tell us what you’re building.
A short message is enough. A senior engineer replies within 1 working day.