Case Study: How I Replaced a $26K/Year Zapier Bill and Built the Lead Operations Engine for a US Marketing Agency

Client: US-based marketing agency (name confidential) Industry: Lead generation for home services (Solar, Roofing, Windows, HVAC, Plumbing, Kitchen, Bath, Siding) Team size: Founder + small ops team (5–10 people) Engagement: 2019 – Present (ongoing, 6+ years)

The situation

The client is a US-based marketing agency that sells home-services leads to buyers across the United States. When a homeowner submits a solar or roofing form, the agency's job is to (a) verify the lead is real, (b) match it to the right buyer based on geography and capacity, (c) deliver it instantly with a recorded call, and (d) get paid for it. Sounds simple. In reality, it's one of the most operationally complex businesses I've ever worked on — every lead touches 6–8 systems before the cash hits the account.

When I joined in 2019, the entire operation ran on Google Sheets.

The problem

Two operations leads were spending most of their day doing things software should have been doing:

Manual lead distribution. Every incoming lead had to be matched to a buyer servicing the right zip code. Multiple buyers often covered the same area, so someone had to manually rotate them, track monthly purchase limits, and route fairly. Mistakes meant unhappy buyers.
Manual call routing. Every lead needed an outbound call from a call center, with the recording attached back to the lead record before the lead could be sold.
Lead quality was a guessing game. Bad leads damaged buyer relationships and triggered refund disputes. The team had no scoring system before delivery.
Onboarding a new buyer took days. Slack channel, payment setup, calendar, CRM, contract — every new buyer meant five tools touched manually.
The previous developer had become unreliable — and the entire revenue engine was sitting on Google Sheets and undocumented Zapier flows.

What I built

Over six years, I rebuilt their operations end-to-end:

1. Lead Distribution Engine (Round-Robin by Zip Code) Custom round-robin algorithm that matches each incoming lead to the right buyer based on zip code coverage, monthly purchase limits, and rotation fairness. Replaced two people doing this manually.

2. Call Center Integration Built a two-way pipeline between the lead system and Vicidial (the call center platform). Leads post to Vicidial automatically; once the agent finishes the call, the recording posts back to the Airtable record — no manual handling.

3. Lead Scoring Pipeline (Custom MVP) Integrated eHawk for fraud and quality verification, then built a scoring layer on top. Every lead gets scored before it reaches a buyer. Trusted Forms (via ActiveProspect) attached for legal compliance on every sold lead.

4. Custom Sales Funnels for 9 Verticals Designed and built funnels for Solar, Windows, Roofing, Kitchen, Bath, Plumbing, HVAC, and Siding — hosted on AWS EC2 with a Namecheap-managed domain layer. Custom JavaScript for auto-completing addresses on every funnel.

5. Solar Heatmap on Thank-You Pages For solar leads specifically, the post-submission page renders a satellite heatmap of the homeowner's actual roof — showing sunlight intensity and projected savings. Built as a buying-intent boost on the thank-you page.

6. Buyer Onboarding Automation A new buyer signing up triggers: Slack channel created, recurring Stripe subscription configured, Calendly synced, GoHighLevel CRM contact created, JotForm intake routed, and DocuSign agreement dispatched — all without human touch.

7. The $26K/Year Zapier-to-AWS Migration This is the one I'm most proud of. The system had grown to 200+ Zapier automations, with the Zapier bill hitting ~$2,200/month. I migrated every workflow to AWS Lambda — today there are 122 Lambda functions running the entire operation. Monthly bill: ~$100. That's a saving of ~$25,000 per year, with better reliability and no task limits.

8. Self-Healing Token Manager Every workflow now reports errors directly to Slack. When GoHighLevel deprecated their old API key system in favor of refresh tokens, instead of patching every workflow, I built a centralized Token Manager — every other Lambda calls it to fetch a fresh token. One change point, system-wide. Anthropic's Claude is wired into the documentation so issues get diagnosed and resolved fast.

Tech stack: Airtable (Vendors, Customers, Zip Codes, Leads), AWS Lambda, AWS EC2, Vicidial, GoHighLevel, Calendly, Acuity, Stripe, Slack, JotForm, DocuSign, ActiveProspect (Trusted Forms), eHawk.

The outcome

~30+ hours per week saved across the operations team. The roles that used to be "manually distribute leads and pull recordings all day" no longer exist as full-time jobs.
~$25,000/year saved on infrastructure by migrating from Zapier to AWS Lambda — without losing a single workflow.
Buyer onboarding went from a multi-day, multi-tool checklist to a single trigger.
Lead quality became measurable for the first time, via the eHawk + scoring layer.
The system is self-monitoring — errors surface in Slack with full context, not silent failures.
6+ years of reliability. The system has scaled with the business, supported new verticals (siding was added recently), and is still running production traffic today.

What this means for your agency

If your operations team is spending hours every week distributing work, copying between tools, or chasing data across spreadsheets — there's almost certainly an Airtable + AWS architecture that can replace 70%+ of that work. And if you're paying four or five figures a month on Zapier or Make, there's likely a six-figure-over-three-years saving sitting in a Lambda migration.

I don't build pretty Airtable bases. I build the operational backbone that lets agencies scale revenue without scaling headcount.

[Want this for your agency? → Book a free 20-minute audit call]