Offline Conversion Tracking

What's included

Capture

• gclid, wbraid, gbraid, msclkid, and fbclid captured on landing and persisted across the session
• First-party cookie or storage setup that survives multi-page journeys and cross-domain hops
• Hidden field population on every form, including third-party embeds and iframes
• UTM values captured alongside the click IDs
• Handling for leads that arrive with no click ID, so they don't break the upload

CRM side

• Fields created and mapped on the lead, contact, and deal object
• Value passed through from deal amount, or a stage-based proxy value where deals close slowly
• Stage mapping so qualified lead, opportunity, and closed won import as separate conversion actions
• Duplicate and merge handling that preserves the original click ID
• Timestamp handling inside the platform import windows

Import

• Google Ads offline conversion import through scheduled upload, API, or a native CRM integration
• Meta offline events or CAPI with lead ID matching
• Microsoft Ads offline conversions where you run it
• Enhanced conversions for leads as a fallback for records with no click ID
• Error monitoring on every upload, because silent import failures are the norm

Deliverable

• End-to-end test from ad click through to the conversion appearing in the platform
• Documentation covering the field map and upload schedule
• Bidding recommendations once the new conversion actions have volume

Why it matters

Lead gen accounts optimize toward whatever you feed them. Feed them form fills and the algorithm goes and finds more people who fill in forms, which is a different group from people who buy. That's how cost per lead trends down for two quarters while the sales team quietly complains about lead quality.

The closed loop fixes it and most of the work is plumbing. Capture the click ID, keep it attached to the record, push the outcome back with a real value. The judgement call is which stage to import and what value to send, because too few conversions and smart bidding starves, too many and you're optimizing on noise again.