MarCloud reports that sync errors appeared in 90% of the portal audits it conducted. That number covers more than just MQLs, but the pattern is familiar to anyone running a HubSpot-to-Salesforce handoff: records disappear, and nobody notices until marketing's pipeline report disagrees with sales' CRM view. Both teams are telling the truth. The leads just never made it across.

Here's what makes this frustrating for ops: "missing MQLs" is not one problem. It's four distinct failure modes wearing the same symptom. Treating them as a single bug guarantees you fix the wrong thing, or create duplicates trying.

Failure Mode 1: The Inclusion Segment

The most common cause is also the most boring. HubSpot's inclusion segment controls which records sync. If an MQL never enters that segment, it never leaves HubSpot. HubSpot's own implementation guidance recommends deciding up front which records should sync and which system is the source of truth for each field. The trouble starts when someone changes the segment criteria without checking whether the segment still catches every MQL.

A subtlety worth calling out: a newly selected segment doesn't push its existing members immediately. Records sync only when a trigger fires, so a contact who qualified before the change can sit there, invisible to Salesforce, until something else updates the record.

Diagnostic: Pull contacts with lifecycle stage = MQL and no Salesforce ID. If that list is long, your segment is the bottleneck. A good default: lifecycle stage is MQL or later, OR Salesforce ID is known (preventing already-synced records from falling out of scope).

Failure Mode 2: Permissions and Access

The integration connects through a specific Salesforce user. If that user lacks object or field-level access, records fail silently. HubSpot's troubleshooting guidance emphasizes checking the sync health dashboard and reviewing record-level error details before changing any settings. Most ops teams skip this and go straight to toggling mappings, which creates a second problem on top of the first.

The fix: audit the integration user's permissions quarterly, and any time a Salesforce admin adds a required field or changes a profile. Set up an error alert so sync failures surface in Slack or email instead of piling up in a dashboard nobody checks.

Failure Mode 3: Salesforce-Side Rejections

This one bites hardest because it looks like a HubSpot problem but lives entirely in Salesforce. Validation rules, duplicate rules, required fields, and restricted picklists can all block a record. Salesforce's documentation on customizing duplicate rules notes that cross-object duplicate matching may require explicit field mapping, and that text values can be truncated when compared across fields with different maximum lengths.

The instinct is to resync. Don't. Resyncing without diagnosing the specific rejection creates repeated failures or duplicates. Start with the error message on the individual record in HubSpot's sync health view.

Failure Mode 4: Routing and Status Mapping

The sneakiest failure, because the record actually synced. It's in Salesforce. But nobody works it.

Two causes. First, HubSpot's lifecycle stage "Marketing Qualified Lead" does not automatically map to Salesforce Lead Status. A record can land as "Open" or "New" instead of anything a rep recognizes as qualified. Second, Salesforce lead assignment rules can route the record to the wrong queue or leave it unowned entirely. Unowned leads are the quietest pipeline leak: they exist in the CRM, count in HubSpot's "synced" total, and nobody is responsible for them.

The Handoff Contract That Prevents All Four

Clean sync settings get MQLs into Salesforce. A handoff contract gets them worked. It needs four elements:

Run It This Week

Setup (30 min): Pull HubSpot contacts at lifecycle stage MQL with no Salesforce ID. Count them. That's your eligibility gap. Then check the sync health dashboard for any error count above zero.

Hypothesis: If we expand the inclusion segment to cover all current MQL qualification paths and clear the top three sync errors by volume, then the MQL-created-to-MQL-received ratio will reach 95%+ within two weeks, because the majority of missing records are eligibility or validation failures, not data loss.

Success metric: MQLs created in HubSpot vs. matching records in Salesforce, measured weekly. Guardrails: duplicate record count in Salesforce stays flat or declines. Stop-loss: if duplicate creation rate increases by more than 10% in the first week, pause and audit the duplicate rules before proceeding.

Test: Create three to four test contacts with realistic data, each qualifying through a different path. Push them to MQL. Confirm each appears in Salesforce within 15 minutes with the correct owner, status, and mapped fields. Delete the test records from both systems.

Most "missing lead" problems aren't data loss. They're configuration decisions producing exactly the outcome they were built to produce, just not the outcome anyone intended. The fix is almost never a resync. It's reading the error message, then reading the contract between the two systems, and noticing where they disagree.