Every rule Ancil runs has been validated in production.
Not documentation. Not best practices. Real portals. Real data. Real failures. 1,500 of them.
Data foundation
The object model determines everything. Properties, lifecycle stages, pipeline stages. Get these wrong and nothing built on top of them is reliable. Ancil audits the foundation first. Always. No exceptions.
- One source of truth per concept
- Lifecycle stage before campaigns
- Required fields gate pipeline advancement
- No duplicate properties across objects
Operational rigor
Every workflow must be explainable, exitable, and reviewable. If you cannot describe what it does, who it targets, and when it stops, Ancil will flag it. No black boxes. No orphaned automation. No workflow that has been running for fourteen months because nobody thought to turn it off.
- Explicit exit criteria on every workflow
- Static lists timestamped or deprecated
- Integrations with owners and kill switches
- Automation documented inside HubSpot, not in a Google Doc nobody reads
Revenue visibility
A report should answer one question. One. If your forecast dashboard pulls from five different pipeline views and nobody agrees on the number, the problem is not your CRM. It is your reporting layer. Ancil flags reports that are doing too many jobs.
- Forecast dashboard is a single view
- Attribution models chosen, not accumulated
- Pipeline reviews use one shared source
- Revenue ops owns the reporting layer
Team fluency
A portal is as healthy as its least-trained user. If a new hire can create any property they want, your data model degrades every quarter. Ancil tracks who is doing what and flags the behaviors that introduce drift.
- Role-based access reviewed quarterly
- Onboarding flows for new team members
- A named HubSpot owner exists and is current
- Training assets linked to the portal, not stored externally
Most audit tools check what HubSpot already flags. Ancil catches what they miss.
Most audit tools check what HubSpot already flags.
The patterns Ancil catches are subtler. The workflow that technically works but will silently inflate your MQL count by 18 percent over six months. The list membership rule that means three thousand contacts are getting the wrong nurture sequence. The property that exists in five objects with slightly different names and is breaking your attribution model.
You only learn to see those patterns by fixing them, repeatedly, at scale.
We have.