Automating Lead Management for a Growth Agency

Problem

GrowthLab Agency, a digital marketing firm, was generating plenty of leads but converting very few. The problem wasn’t lead quality — it was speed. Leads came from 8 different sources (web forms, Facebook ads, LinkedIn, cold emails, referrals, webinars, trade shows, and phone calls), and each had to be manually entered into HubSpot.

By the time a lead was in the CRM and assigned to a sales rep, days had passed. In the agency business, speed-to-lead is critical — research shows that responding within 1 hour makes you 7x more likely to qualify the lead. GrowthLab was responding in 2-3 days.

The manual process also meant data was inconsistent. Some leads had phone numbers but no email, others had company names but no industry. Deduplication was non-existent, so the same lead often appeared multiple times.

Solution

I built a multi-source lead automation pipeline using n8n as the orchestration layer. The system handles every lead source through a single, standardized pipeline.

Architecture

The pipeline follows these stages:

1. Capture: Webhooks from all 8 lead sources feed into n8n workflows. Each source has a dedicated webhook endpoint.
2. Normalization: A custom n8n function node standardizes all data — phone numbers are formatted, company names are cleaned, email addresses are validated.
3. Deduplication: The system checks HubSpot for existing contacts with the same email or phone number. Duplicates are merged, not duplicated.
4. Enrichment: Clearbit API enriches each lead with company size, industry, revenue, and tech stack.
5. Scoring: A custom scoring algorithm assigns a score from 1-100 based on source quality, company size, engagement level, and budget indicators.
6. Routing: Leads scoring 70+ are immediately assigned to senior reps and trigger a Slack notification. Leads scoring 40-69 go to the general queue. Leads below 40 enter a nurture sequence.
7. Follow-up: Each lead enters an automated email sequence appropriate to their score and source.

Key Design Decisions

• n8n over Zapier: While Zapier would have been simpler for some integrations, n8n’s self-hosted nature and lower cost at scale made it the better choice. The ability to write custom JavaScript in function nodes was also critical for the scoring algorithm.
• Real-time over batch: Every lead is processed immediately, not in batches. This ensures speed-to-lead is always under 1 minute.
• Human override: Sales reps can adjust lead scores and routing rules through a simple admin interface, so the system improves over time based on their feedback.

Architecture

8 Lead Sources → Webhooks → n8n Workflows
                                ↓
                        Data Normalization
                                ↓
                        Deduplication (HubSpot)
                                ↓
                        Enrichment (Clearbit)
                                ↓
                        Lead Scoring Algorithm
                        ↙        ↓        ↘
                  Senior Rep   General    Nurture
                  + Slack      Queue     Sequence
                      ↓          ↓          ↓
                   HubSpot CRM + Email Automation

Technologies

• n8n: Workflow orchestration and automation logic
• HubSpot API: CRM for lead storage and management
• Clearbit API: Company data enrichment
• Slack API: Real-time notifications for high-priority leads
• SendGrid: Automated email sequences
• PostgreSQL: Lead scoring rules and configuration

Results

• 100% of leads are now captured and entered into the CRM within 1 minute
• 3x improvement in lead-to-customer conversion rate
• Zero leads are lost due to manual entry delays
• 20+ hours per week saved across the sales team
• Sales reps now spend their time selling, not doing data entry

Lessons Learned

1. Speed matters more than perfection: Getting a lead into the CRM in 1 minute with 90% data accuracy is better than getting it in there in 2 days with 100% accuracy. You can always enrich later.
2. Scoring is iterative: The initial scoring algorithm was too aggressive — too many leads were routed to nurture. We adjusted the thresholds based on actual conversion data over the first month.
3. Source quality varies enormously: Leads from referrals converted at 45%, while leads from Facebook ads converted at 8%. The scoring algorithm needed to weight source heavily.
4. Sales team buy-in is critical: The system only works if sales reps trust it. We involved them in designing the scoring criteria and routing rules, which made all the difference.
5. Monitor for source changes: Lead sources change over time — a webinar series might produce great leads one month and terrible leads the next. The scoring system needs regular recalibration.