Case Study: How [Brokerage] Doubled Leads Using CRM Automation

Lead generation is a numbers game. [Brokerage] was playing it with an abacus. They had three primary lead sources: a Zillow Premier Agent account, their main website contact form, and manual entries from open houses. Each source fed leads into a shared inbox, creating a digital free-for-all where agents would cherry-pick promising inquiries and leave the rest to rot.

The result was a lead response time measured in hours, not minutes. By the time an agent finally called a prospect from a web form, that prospect had already spoken to three other realtors. Their CRM was a graveyard of stale data. They weren’t losing leads. They were actively killing them.

The Diagnosis: A Leaky Bucket Architecture

The core problem wasn’t the quantity or quality of the leads. It was the complete absence of a system to process them. The manual workflow was the bottleneck. An agent would have to see the email, copy the contact info, paste it into the CRM, create a new deal, and then maybe, if they weren’t distracted, make a call.

This process had a failure rate north of 40%. Leads were missed, data was entered incorrectly, and there was zero accountability. You can’t build a predictable sales pipeline on that kind of foundation. It’s like trying to move water in a sieve. The effort is there, but the output is pathetic.

We identified three critical failure points:

• No Central Intake: Leads arrived via different protocols (Zillow API push, SMTP from the web form, CSV uploads). There was no single point of truth.
• Manual Assignment: The “fastest gun in the west” approach to lead claiming created internal friction and left less desirable, but still valuable, leads to go cold.
• Delayed Follow-up: The time gap between lead submission and first contact was the single biggest killer of conversion.

The Solution: Building an Automated Lead Router

We needed to bypass the shared inbox entirely. The goal was to build a machine that could catch, clean, route, and log every single lead within seconds of its creation. We decided to use n8n as the middleware brain, connecting their lead sources directly to their Pipedrive CRM.

The architecture was straightforward. All lead sources would be funneled to a single n8n webhook URL. Zillow was configured to POST new leads there. The website form was modified to do the same. Manual entries were handled by a simple Google Sheet connected to the workflow, which was a necessary evil for handling Luddites at open houses.

Once a lead hits the webhook, the workflow executes a series of steps:

1. Data Standardization: Zillow’s data structure is different from a simple web form. The first node in our workflow hammers the incoming JSON into a consistent format. We strip junk characters, format phone numbers to E.164, and normalize address fields. You can’t trust incoming data. Ever.
2. Duplicate Check: We execute an API call to Pipedrive to search for an existing person by email or phone number. If a match exists, we append the new inquiry as a note on the existing contact instead of creating a duplicate. This keeps the CRM clean.
3. Round-Robin Assignment: This was the key to fairness and speed. We maintained a simple list of active agent IDs in a static data node. A bit of simple code routes the lead to the next agent in line. This isn’t rocket science, it’s just logic.

// Simple expression in n8n's "Set" node to get the next agent ID
// Assumes agentIds is a static array like [101, 102, 103]
{{ 
    const agentIds = [101, 102, 103];
    const executionIndex = $execution.id; // A pseudo-random number for distribution
    const agentIndex = executionIndex % agentIds.length;
    return agentIds[agentIndex]; 
}}
    

The code snippet above shows the logic for assigning an agent. It uses the execution index, a reasonably random number, with a modulo operator to cycle through the array of agent IDs. It’s cheap, effective, and requires zero database lookups.

Creating the Deal and Firing the Alerts

With the data cleaned and an owner assigned, the final step is to push the information into the CRM and alert the agent. The workflow makes a `POST` request to the Pipedrive API to create a new “Person” and a new “Deal” simultaneously, linking them together.

The API payload has to be precise. The CRM’s documentation claimed certain fields were optional, but we found that omitting them caused the deal to be created without being visible on the agent’s dashboard. A classic “documentation lie” that cost us a few hours of debugging.

{
  "title": "New Lead: {{ $json.body.firstName }} {{ $json.body.lastName }}",
  "person_id": {{ $items("Create Person").first().json.data.id }},
  "user_id": {{ $items("Set Agent ID").first().json.agentId }},
  "stage_id": 1,
  "custom_fields": {
    "lead_source": "{{ $json.body.source }}",
    "property_inquiry": "{{ $json.body.propertyAddress }}"
  }
}
    

Immediately after the API call confirms success, the workflow fires off two alerts. One is a Slack message posted to a dedicated #new-leads channel, creating public accountability. The second is a direct SMS to the assigned agent’s phone with the lead’s name and number. The expectation is clear: call this person now.

The trade-off here is cost. Every SMS sent is a few cents. If you have a flood of low-quality or spam leads, this automation becomes a wallet-drainer. We had to heavily fortify their website form with a honeypot and reCAPTCHA to ensure we weren’t just texting bots all day.

The Results: More Than Just “Doubled Leads”

The headline is that they “doubled their leads.” The reality is more specific. Their raw lead volume stayed the same. What doubled was their lead-to-qualified-opportunity conversion rate. They were closing twice as many deals from the same pool of inquiries because they finally stopped dropping the ball.

The numbers don’t lie:

• Average Lead Response Time: Dropped from over 4 hours to under 5 minutes.
• Lead Leakage Rate: Went from an estimated 40% of leads never being contacted to less than 1%.
• Agent Productivity: Freed up nearly an hour per day per agent from mind-numbing data entry. Morale improved because the system was seen as fair.

The old process was like trying to shove a firehose through a needle. Gallons of potential revenue were spraying everywhere except where they needed to go. The automation didn’t create more water, it just attached a proper nozzle, directing the full force of their lead flow into their sales pipeline.

This wasn’t a complex, multi-month project. We built and deployed the core logic in under a week. The biggest hurdle wasn’t the technology, it was getting the agents to trust the machine. Once they saw high-quality leads appearing on their dashboard with all the necessary information, they became converts. They stopped fighting for scraps in the inbox and started focusing on what they were paid to do: sell.