When your HubSpot account has dozens of forms but no standardized lead type field, most teams end up routing leads manually — or not routing them at all. The fix is simpler than it sounds: use the form name as a classification signal, pass it to an AI, and let the automation route each lead into the right workflow.
The Problem: Unstructured Leads, Disconnected Prompts
HubSpot accounts accumulate forms over time — one per campaign, landing page, webinar, and product offering. The form name implicitly tells you a lot about a lead (is this from a training campaign? a product inquiry? a partnership request?), but that context rarely gets surfaced in the CRM in a way automation can act on.
Without a "lead type" or "lead source category" field that's populated consistently, conditional workflows break down. You end up with a flat list of leads and a library of follow-up prompts written for different audiences, with no reliable way to match them automatically.
A related version of this problem: the team has a large, unstructured prompt library — sometimes shared across email threads — where different prompts apply to different lead types, but no one has formally categorized which prompt goes with which scenario.
The Solution: AI Classification via Form Name
The approach, using Zapier (or Make), works in five steps:
- Trigger on a new contact or lead created in HubSpot
- Pull the form name from the HubSpot activity or contact properties
- Pass the form name to an AI step (OpenAI, Claude, or Zapier's built-in AI) with a classifier prompt
- The AI maps the form name to a lead category based on keyword patterns you define (e.g., forms containing "training" → Training Lead; forms containing "inquiry" → Sales Lead)
- Write the category back to HubSpot as a contact property, then branch the workflow accordingly
The result: every new lead gets categorized automatically, and the right follow-up sequence fires without any manual tagging.
Handling a Messy Prompt Library
If your team's follow-up strategy lives in an unstructured collection of prompts — scattered across documents or email threads — start with a one-time AI clean-up pass before building the automation:
- Feed the full prompt document to an AI and ask it to identify where the breaks are between individual prompts and infer what each one is for. This produces a categorized prompt map.
- With the map in hand, your Zapier or Make automation can use the form name to select the right prompt category, then apply it to the incoming lead.
This turns a chaotic collection of copy into a decision tree that automation can traverse cleanly.
What to Prepare Before You Build
Before wiring up the automation, do a one-time exercise with your team:
- List all active HubSpot forms
- Group them by lead intent (Sales, Training, Support, Partnership, etc.)
- Provide those groupings to the AI classifier as context in the system prompt
With that groundwork done, the automation classifies new leads accurately — even as new forms are added over time, as long as the naming patterns stay consistent.
Dealing With Non-Technical Stakeholders
A recurring challenge in projects like this is that the business owner understands their lead types intuitively but struggles to articulate the rules in a way that's actionable for automation. A few approaches that help:
- Show, don't ask: Present a sample of 10–15 leads and ask the stakeholder to sort them into categories manually. That session usually surfaces the classification rules faster than any meeting.
- Start with keywords: Even a rough list of 3–5 keywords per category is enough to build a working classifier. Refine it after the first real batch runs.
- Use the activity field as a fallback: If the form name isn't available, HubSpot's activity log often contains form submission data that can serve the same purpose.
Automate Your HubSpot Lead Workflows
If you want help building a lead classification and routing system that connects HubSpot, AI, and your existing tools, our automation team can design and implement it in a single engagement.
