A CRM data quality audit is seven checks you can run in an afternoon: duplicates, forecast fields, freshness, past close dates, owners, source and one agreed pipeline number. Each has a pass mark. Pass six or seven and sales can trust the numbers; pass three or fewer and nobody should forecast from the CRM alone.
Most leadership teams already suspect their CRM is wrong. Few can say how wrong, or where. A vague sense that “the data is bad” leads to a vague clean-up, which decays again within months.
An audit replaces the suspicion with seven numbers. Each number points to one fix, and the fixes have an order. This guide gives you the checks, the pass mark we use for each, a simple way to score the result and the order we fix things in. If you want to see the problem first, our live teardown, Your CRM is lying to you, shows five of these checks run on screen.
Why do sales teams stop trusting CRM numbers?
Because the people closest to the data can see how little of it holds up. In Salesforce’s 2024 State of Sales research, only 31% of sales professionals in the UK and Ireland said they fully trust the accuracy of their organisation’s data. The survey covered 5,500 sales professionals in 27 countries between March and April 2024.
The people who run CRMs say the same. In Validity’s 2025 survey of 602 CRM users and administrators in the US, the UK and Australia, 76% said less than half of their CRM data is accurate and complete. 37% said their company loses revenue as a direct result.
The confidence gap is older than AI. In Validity’s 2022 survey of 1,241 CRM administrators, 76% rated their data quality as good or very good. Yet in the same survey, 44% estimated that poor CRM data costs their company more than 10% of annual revenue.
In our experience the cause is rarely one big failure. It is a few small habits: deals nobody updates, close dates that slip without anyone moving them, two records for the same buyer, and a pipeline number that marketing, sales and finance each define differently. Each habit is measurable, which is what the seven checks below do.
What are the seven checks in a CRM data quality audit?
Run them on open deals and on the accounts and contacts your team works today, not on the whole archive. Most CRMs can answer each one with a saved report or list view; none needs a new tool. The pass marks are the thresholds we use with clients, not an industry standard. Set your own if you have a reason to.
1. Duplicates: is each buyer one record?
Duplicates split the history of an account across records, so nobody sees the whole relationship and two reps can chase the same buyer. Export companies and match on website domain; export contacts and match on email address. Count the records that share a domain or an email with another record.
Pass mark: fewer than 5% of active company records are duplicates.
2. Forecast fields: are the five fields the forecast needs filled in?
A forecast is only as good as the fields under it. Pick the five it depends on, usually stage, amount, close date, owner and next step. Filter open deals for any of the five left blank.
Pass mark: at least 90% of open deals have all five filled in.
3. Freshness: has anyone touched the deal recently?
A deal nobody has updated is a guess, however complete it looks. Filter open deals by last activity or last modified date, then by time in the current stage.
Pass mark: at least 80% of open deals updated in the last 30 days, and none in the same stage for more than 60 days without a logged reason.
4. Past close dates: does the pipeline admit what slipped?
An open deal with a close date in the past is the clearest sign of a pipeline nobody maintains. It also inflates the current quarter. Filter open deals where the close date is before today.
Pass mark: fewer than 5% of open deals.
5. Owners: does every record have a person who answers for it?
A record nobody owns is a record nobody fixes. Look for accounts, contacts and deals with no owner, and for records owned by people who have left, which many CRMs still show as active.
Pass mark: zero open deals and zero target accounts without an active owner.
6. Source: can you trace a won deal back to where it began?
Without a source, every budget decision about marketing is a guess. Take the deals won in the last two quarters and check the original source field, and whether it uses your agreed list of values rather than free text.
Pass mark: at least 90% of won deals have a source from the agreed list.
7. One number: do marketing, sales and finance report the same pipeline?
This is the check that decides trust. Ask each team for this quarter’s open pipeline and its definition: which stages count, which dates, which currency. Put the three numbers side by side.
Pass mark: one written definition, and the three numbers agree within 5%.
How do you score the audit?
Give one point for each check you pass. The total tells you how far the numbers can be trusted today, and what to do this quarter.
| Score | What it means | What to do |
|---|---|---|
| 6 or 7 | Sales can trust the numbers. | Keep the weekly review and rerun the audit each quarter. |
| 4 or 5 | The forecast is usable with care. | Fix the failed checks in the order below before the next quarterly review. |
| 3 or fewer | The CRM describes the past, not the pipeline. | Do not forecast from the CRM alone. Fix ownership and the forecast fields first. |
What should you fix first?
Fix in the order that stops new damage before you clean old damage. A clean-up without the first two steps decays again within months, because the same habits keep producing the same errors.
- Owners (check 5). Give every open deal and target account an active owner, and one person accountable for data quality overall, with authority to change how the CRM is used.
- Rules at the door (checks 2 and 6). Make the five forecast fields and the source required when a deal is created, and turn free-text fields into picklists.
- The one number (check 7). Write down the pipeline definition and get marketing, sales and finance to sign it.
- The backlog (checks 3 and 4). Ask each owner to update or close their stale deals and past close dates within two weeks.
- Duplicates (check 1). Merge last, once owners exist to decide which record wins.
Then make it a habit. Put the seven numbers on one screen and review them weekly with sales, marketing and RevOps together. The audit becomes a dashboard, and trust becomes something you can see moving.
If the audit shows the problem goes beyond the CRM, our guide to auditing a martech stack covers the tools around it. All our free tools are on the Diagnostics page.
How does your CRM data compare with the rest of your revenue engine? The Revenue Engine Health Check scores your revenue data alongside four other areas, and shows which one to fix first. Five minutes, free.
Frequently asked questions
How long does a CRM data quality audit take?
An afternoon for most mid-sized companies. Each of the seven checks is one report or list view in the CRM. Check 7 takes longest, because it needs three teams to share their numbers and definitions.
How often should we audit CRM data quality?
Run the full audit once a quarter, before the quarterly business review. Between audits, review the seven numbers weekly so problems surface while they are still small.
Who should own CRM data quality?
One named person, usually in revenue operations, with the authority to change required fields and how the CRM is used. In Validity’s 2025 survey, 46% of companies had nobody working full time on CRM data quality.
Do we need a data quality tool to run the audit?
No. Every check runs on standard CRM reports or an export to a spreadsheet. Tools help to keep data clean once you know which checks fail; they do not tell you what “good” means for your pipeline.
What is the difference between a CRM audit and a data hygiene clean-up?
The audit measures; the clean-up fixes. Run the audit first, so the clean-up targets the checks that failed and you can show the improvement afterwards.
Sources
- Salesforce, State of Sales research, UK and Ireland results, 2024 (5,500 sales professionals in 27 countries, surveyed 8 March to 18 April 2024)
- Validity, The State of CRM Data Management in 2025, as reported by MediaPost, 11 July 2025
- Validity, 2022 State of CRM Data Health, press release, 10 February 2022