Convert ADF XML leads into HubSpot with Zapier

What ADF is (and why it breaks in Zapier)

The reliable ADF to HubSpot fix: XML → JSON inside the Zap

1. Trigger: New email arrives with the ADF XML in the body
2. Code step: Convert XML to JSON
3. HubSpot: Find or create Contact
4. HubSpot: Find or create Deal (optional)
5. HubSpot: Update properties using the mapped fields

Step-by-step build (email → HubSpot)

1) Trigger on the inbound ADF email

• A dedicated inbox address that only receives ADF leads
• A label/folder rule that routes only lead emails into the Zap trigger

2) Add a Code step to convert XML to JSON

from defusedxml import ElementTree as ET

xml_text = input_data.get("xml", "").strip()
if not xml_text or len(xml_text.encode("utf-8")) > 100_000:
    raise ValueError("Missing XML or payload exceeds this example limit")
root = ET.fromstring(xml_text, forbid_dtd=True)
if root.tag != "adf":
    raise ValueError("Expected a non-namespaced adf root")
prospects = root.findall("prospect")
if len(prospects) != 1:
    raise ValueError("This example requires exactly one prospect")
p = prospects[0]
contact = p.find("customer/contact")
if contact is None:
    raise ValueError("Missing customer/contact")

def text(node, path):
    return (node.findtext(path) or "").strip()

names = {
    n.get("part"): (n.text or "").strip()
    for n in contact.findall("name")
}
output = {
    "lead_id": text(p, "id"),
    "first_name": names.get("first", ""),
    "last_name": names.get("last", ""),
    "email": text(contact, "email"),
    "phone": text(contact, "phone"),
    "vehicle_vin": text(p, "vehicle/vin"),
    "vehicle_year": text(p, "vehicle/year"),
    "vehicle_make": text(p, "vehicle/make"),
    "vehicle_model": text(p, "vehicle/model"),
    "provider": text(p, "provider/name"),
    "request_date": text(p, "requestdate"),
}
output["needs_review"] = not bool(output["email"])

3) Map fields into HubSpot Contact

• Email → Contact email
• First/last name → Contact name fields
• Phone → Contact phone
• Lead source → a custom property (or lifecycle fields, depending on your setup)

4) Create or update a Deal (if your pipeline needs it)

• Deal name convention (e.g., “New Lead — {{Last Name}}, {{Vehicle Model}}”)
• Pipeline/stage defaults
• Owner assignment rules (round-robin, territory, or fixed owner)

5) Add guardrails so it doesn’t fail silently

• If parsing fails, retain the payload in restricted storage and alert the owner with a message ID and failure reason. Do not broadcast customer details or the complete raw email to a shared channel.
• Use a Filter or Path for needs_review. Missing or invalid contact data should go to a review queue, not blindly create an incomplete HubSpot record.
• Normalize contact identifiers and check existing records before creating them. A shared phone number is not proof that two people are the same.
• Deduplicate the inquiry separately from the contact, using the provider and stable lead ID when available, plus the source email message ID for replay protection.
• Store the resulting HubSpot IDs. Retrying an email should not create a second deal or re-trigger notifications.

Example field mapping cheat sheet

• Customer
• Name
• Email
• Phone
• Address
• Vehicle
• Year, make, model, trim
• VIN (when present)
• Lead metadata
• Provider/source
• Received timestamp

When to keep it in Zapier vs. move to a dedicated parser

• The lead volume is high enough that occasional manual fixes are too costly
• You have multiple lead formats to support (not just ADF XML)
• You need advanced validation or enrichment before the CRM step

Get help mapping your ADF leads into HubSpot