A dirty CRM is not a data problem. It's a business problem. Every decision your team makes downstream — forecast accuracy, marketing attribution, customer health scores, renewal timing — depends on what's in the CRM. When the data is wrong, every decision built on top of it is wrong too.

I've audited dozens of CRM setups. The most common issues are almost always the same: missing required fields, duplicate records, stage definitions that mean different things to different reps, and historical data that never got migrated or updated correctly. None of it is catastrophic on its own. Together, it compounds into a pipeline your leadership team doesn't trust.

Here's how to fix it.

Phase 1: Audit before you clean

The most common mistake businesses make when cleaning a CRM is starting with cleanup before they understand what's actually broken. You end up spending three days fixing issues that weren't the real problem.

Before touching a record, run these five checks:

01
Pull a duplicate report
In HubSpot, use the Duplicates tool under Contacts or Companies. In Salesforce, use the Duplicate Management rules or a third-party tool like Dedupely. You're looking for any records where two or more represent the same company or person. Note the volume — more than 5% duplication is a red flag.
02
Check your required field completion rate
Export a list of all open deals. Look at how many are missing critical fields: company name, deal owner, close date, deal stage, lead source. Any field that is missing on more than 15% of open deals is a systemic problem, not a rep problem.
03
Review your pipeline stages
List every pipeline stage you have. For each one, write down what has to be true for a deal to be in that stage. If you cannot write it down, your reps cannot apply it consistently. Stages without exit criteria are a forecast liability.
04
Find stale deals
Flag any open deal with no activity in 30 days. These are either dead deals that were never closed out (inflating your pipeline) or real deals that have gone dark and need attention. Either way, they need to be addressed.
05
Map your lead sources
Pull a count of records by lead source. Look for blanks, "unknown," and variations on the same source (Web, Website, web form, Webform). Inconsistent lead source data makes attribution impossible and marketing ROI invisible.

Phase 2: Clean systematically, not randomly

Now that you know what's broken, clean in this order:

  1. Merge duplicates — start with contacts attached to open deals, then companies, then the rest.
  2. Close out dead deals — create a "Lost" or "Disqualified" stage and move stale deals there with a reason. Get them out of your active pipeline.
  3. Standardize field values — pick one naming convention for lead sources, industry types, and close reason categories. Use picklist fields instead of text wherever possible.
  4. Fill required fields on active deals — set aside time with each rep to update open records. Make it a sprint, not an ongoing ask.
One rule to follow: Never delete records. Archive them. A "deleted" contact usually means you lose the activity history. Mark records as disqualified, churned, or inactive — and then filter them out of your views. You can always surface old data. You can't un-delete it.

Phase 3: Lock it down so it doesn't break again

Cleanup without prevention just means you're doing the same project again in 12 months. Once the data is clean, these are the guardrails that keep it that way:

How long does a CRM cleanup actually take?

For a business with under 5,000 contacts and under 200 open deals, a thorough cleanup typically takes two to four weeks of focused work. The audit takes a few days. The actual cleanup takes another week. Locking it down and training the team takes a few more days.

The businesses that take six months are the ones that try to clean and keep running business as usual at the same time. If you're serious about fixing your CRM, you need to temporarily slow down how many new records are being created while you clean the existing ones — or at minimum, freeze any field-level changes until the cleanup is complete.

Want a second opinion on your CRM setup?

I'll spend 30 minutes looking at your actual setup and tell you exactly what's broken and what to fix first. No proposal, no pitch deck — just a straight answer.

Book a Quick Review →

If you're not sure where to start, the RevOps Scorecard includes a data quality section that will surface your biggest gap in about three minutes. Take it free here.