A sample of 10 to 25 mixed rows can reveal missing owners, unclear next actions, possible duplicates, stale statuses, missed calls, and rows that should be held. Do not choose only the cleanest records. Include a few different sources and owners, and include at least one row where the next decision is unclear.
For a safe CRM sample for a first review, ask three questions: What event created the row? What evidence supports the next action? What would stop another customer-facing message? If the sample does not contain enough information to answer those questions, mark the row Missing Context instead of adding more private data by default.
The Missed Lead Recovery review accepts a small redacted browser sample. It does not require a CRM password, send messages, change records, or decide contact permission. The owner reviews each category and decides whether a broader review is appropriate.
Privacy is part of a useful cleanup. A smaller sample with clear boundaries is easier to explain, easier to delete or replace, and less likely to expose information that the review never needed. The first result should be a repair list and a set of open questions, not a larger private copy of the business.
Start with a bounded review: use a small redacted sample. AI Cleanup Doctor does not send messages, change a CRM, or decide contact permission. See the local-only review.
Review boundary: This article organizes supplied evidence. It does not prove platform fault, employee fault, attribution accuracy, duplicate billing, consent, lead quality, calls, jobs, rankings, orders, ROI, revenue or AI citations.