A new VP of Marketing inherits the automation platform, the CRM, the connector between them, and a team that has stopped trusting all three. The temptation is to pick a rebuild and start. The audit exists so that the first decision is made on evidence rather than on whoever tells the most confident story in week one.
Everything below is read-only. Nothing is changed. Every finding is tied to an export the client can re-run.
The ten surfaces
- Object inventory and surplus rate. Contacts, companies, deals, custom objects: counts, growth by month, and the share that is surplus by a stated definition (same domain, same key, same normalised name). The number is less important than the trend.
- Duplicate provenance. Not how many, but which mechanism made them. In the case study, records carrying a system-of-record key were 99.7% clean and nine in ten surplus records were minted by the CRM itself — two events accounted for most of it. Provenance is what tells you whether to fix a sync or an import standard.
- Lifecycle and stage definitions. What each stage means in each system, who sets it, whether the two systems agree, and how often a record moves backwards. Disagreement here is the usual root of an untrusted funnel.
- Handoff integrity. Where leads go when marketing is done with them, how many arrive attached to the wrong account, and how long routing takes. The case-study portal was failing a large share of handoffs because contacts hung off the wrong one of several identical company shells.
- Workflow and program history. Every active automation, what it writes, what it creates, and the twelve-month record of what it has done. This is where the “list-upload workflow that created tens of thousands of companies in one day” is found.
- Import history. Who imported what, with which match key, and what each import created versus updated. A contact import with no domain column is a company-creation engine.
- Connector configuration. Marketo–Salesforce, Pardot–Salesforce, HubSpot–Dynamics: field-level ownership, sync direction, the error queue and its age, and the rules nobody documented when the person who set them left.
- Property and field hygiene. Fill rates on the fields the reporting depends on, orphaned properties, duplicated definitions across systems, and the free-text fields doing the job of a picklist.
- Attribution and reporting spine. What the revenue dashboard actually reads from, whether closed-won reconciles to finance, and where the last-touch flattery lives.
- Access and credentials. Integration users, API keys, who can bulk-edit, and whether any credential leaves the platform. The audit doesn’t change any of it; it lists it.
What comes out
A scored findings deck, evidence exports for every score, a recorded walkthrough, and a scope memo: what a fix requires, the route the platform gives you, the guards it needs, the order it has to run in, and what it costs. Where the honest answer is “don’t touch this yet”, that is the recommendation.
The audit is three weeks, USD $4,500 up to 100k records, and credited in full against a Systems Rescue Sprint or Custom Build within 30 days. If a first call shows you don’t need one, I’ll say so.