The forecast looks fine until someone starts asking about the deals.
One customer stopped responding weeks ago. Another opportunity has moved into next month for the third time. A salesperson left, but their open deals are still sitting under their name. Someone says a sale is missing, and someone else opens a spreadsheet to explain why the CRM total is wrong.
The meeting was supposed to be about what to do next. Now it is about which number to believe.
That is a good reason to run a CRM audit. Before making another forecast or asking for more leads, check whether the records describe the business you are actually running. This CRM audit checklist gives you eight places to start.
What is a CRM audit?
A CRM audit checks the customer records, sales opportunities, and reporting your team relies on. Are the right records there? Do the stages mean anything? Can someone explain what happens next with each open deal?
A useful audit starts with a question the business needs answered. Can we trust next month's forecast? Why do two stores report such different close rates? Why does the sales total change depending on who pulls the report?
For this review, start with the pipeline. Follow a set of records from the original inquiry through the latest conversation or sale. See how well the system's version holds up.
How to audit CRM data quality before the next sales review
Pick one pipeline and a defined period. Save a copy of the records before changing anything. You will want the source, owner, stage, amount, close date, last customer contact, next action, and outcome, along with record IDs so you can trace each deal.
Use the full export to count obvious gaps, then open individual records. Fifty to a hundred can be a manageable first pass, depending on the size and variety of the business. Include different sources, stages, stores, and outcomes. Keep any problem records you deliberately picked separate from the broader sample; otherwise, your results will reflect what you went looking for.
Read the notes. Check the conversations. Match a few recorded sales to the transaction records. This is where a field that looks complete can turn out to be wrong.
IBM treats completeness and accuracy as separate measures of data quality. You can see why: a close date filled in automatically gives you a value. Whether the buyer expects to close that day is another question.
CRM audit checklist: eight checks
- Did every real inquiry make it in?
- Are you counting the same deal twice?
- What does each sales stage actually mean?
- Has the customer moved, or just the record?
- What supports the close date and amount?
- Who owns the next move?
- Do the recorded sales match what sold?
- Are the stores counting the same way?
1. Did every real inquiry make it into the CRM?
Open the website inquiry log, phone records, or chat history and compare it with what arrived in the CRM. Match records by IDs where you can. Leave uncertain matches on a list to investigate.
Some missing entries will have a reasonable explanation, such as spam or an existing customer calling about service. Others may reveal a form that stopped syncing or a phone inquiry someone meant to enter later. Find out which is which.
The CRM cannot tell you about an inquiry it never received. Count those gaps before judging how well the team worked the leads. Once arrival is confirmed, the speed-to-lead diagnostic helps trace what happened after assignment.
2. Are you counting the same deal twice?
A customer submits a form, calls the store, and comes back through a different website. Check whether those touches belong to one purchase or several. Three records can make the pipeline look larger while leaving three people with different pieces of the conversation.
Repeated contact details are a starting point. People can share a phone number, and one customer can have more than one legitimate deal. Read the history before merging anything, preserve a copy, and use the CRM's supported process. Salesforce's duplicate-management guidance covers the available tools.
Once you confirm a duplicate, look at how it got created. Cleaning up this month's list leaves next month's problem intact if the same entry path keeps creating extra records.
3. What does each sales stage actually mean?
Ask two managers what qualifies as an appointment. One may mean a confirmed day and time. The other may mean the customer said they might come by Saturday. Both stores have a number in the same column, but they are counting different commitments.
Go stage by stage and agree on what has to happen before a deal moves forward. Then open records in each stage and look for that evidence. An appointment might need a confirmed time; a proposal might need to have reached the buyer.
Keep the rules simple enough for the team to use. A stage should help the next person understand where the customer stands without having to ask the salesperson to translate it.
4. Has the customer moved, or just the record?
A deal was updated this morning. The customer has not answered in three weeks. Both can be true.
A workflow, import, or internal edit can refresh a record without a customer conversation. HubSpot separates last modified, last activity, last contacted, and next activity dates. Check which one your report uses and what counts toward it.
Look at the last meaningful exchange, time in stage, and the next agreed action together. A long sale can still be moving. A recently edited deal can be going nowhere. Use your own sales cycle to decide what deserves a closer look.
If you report on days in stage, check how the CRM calculates them. A deal that leaves a stage and returns may be counted differently across views, as Salesforce's stage-duration guide explains.
5. What supports the close date and amount?
Pull the open deals whose close date has already passed. Then look at the ones that keep moving into the next month. Ask what the current date is based on. Did the buyer agree to a decision date? Is financing pending? Is someone waiting on a contract?
Check the settings, too. HubSpot can fill an unset close date with the last day of the month or a configured default. A pile of month-end deals may tell you something about the setup before it tells you anything about the sales team.
Do the same with the amount. Make sure everyone is using the same currency and value: quoted revenue, expected gross profit, and collected payment answer different questions. The forecast needs a basis the people in the meeting understand.
6. Who owns the next move?
Open a deal and ask who is responsible for it today. Then ask that person what happens next. The answers should agree.
Watch for former employees, shared queues nobody checks, and records assigned to a manager who assumes a salesperson is handling them. An owner field can be filled in while the work itself belongs to nobody.
“Follow up” needs a little more detail. “Confirm the buyer's decision after Tuesday's finance review” gives the owner a purpose and a date. Agree who covers it if that person is out. These are the handoffs where responsibility can disappear even when the system looks organized.
7. Do the recorded sales match what sold?
Take the deals marked won and match them to the record your business uses to confirm a sale. That might be a signed agreement, booked job, invoice, or delivered vehicle. Be clear about the event: a signed contract and a collected payment can fall in different months.
For a dealership, compare CRM outcomes with the agreed sold or delivered event in the dealer management system (DMS). Use the same date rules and matching IDs, then check both directions. Which won deals have no transaction? Which transactions have no corresponding CRM outcome?
Some differences will be timing. Others may be cancellations, duplicates, or outcomes nobody updated. Write down the explanation so the next meeting does not spend another twenty minutes working through the same discrepancy.
8. Are the stores counting the same way?
Before ranking stores, compare the reports they use. Look at date ranges, time zones, excluded sources, duplicate handling, and whether each report counts people, inquiries, or opportunities. A difference in a filter can become a conversation about performance that the data does not support.
Close rates need the same attention. Dividing this month's sales by this month's new leads may mix buyers who arrived months apart. When you want to know what happened to a group of leads, follow that same group through an appropriate sales window.
Agree on the definitions before deciding which store has a sales problem. That is the discipline behind a useful performance review.
Give the findings somewhere to go
By now, you should have a short list of records that need attention and a clearer idea of why. Give each issue an owner, an action, and a date to check back. Keep the confirmed errors separate from the deals someone still needs to investigate.
A working example (invented figures)
You review 50 open opportunities. Twelve have close dates that have already passed, and nobody has checked whether the buyers still intend to move forward.
The sales manager assigns a review of those deals. The CRM administrator checks how the dates were set. At the next meeting, the team brings back what it learned, what changed, and which deals still need an answer.
Those twelve records need a review. Calling them lost revenue would get ahead of the evidence. The point is to find out which deals still belong in the forecast and what needs to happen with each one.
Start with the issues affecting a decision you need to make now or leaving customer work unattended. Then fix the rule, integration, or habit that created them. Gartner's data quality guidance puts business priorities and ownership at the center of the work. That is useful advice for a CRM cleanup, too: make the fix matter to the people running the business.
How often should you audit your CRM?
Work the open issues into the sales meeting you already have. A monthly check of duplicates, stale deals, ownership, and recorded outcomes can be a reasonable starting point. Higher volume or a faster sales cycle may call for more frequent checks.
Take another look after a migration, integration change, or new store launch. Pay attention to fresh records as well as the backlog. If the same problem keeps appearing, the cleanup has not reached the cause yet.
Get back to talking about the business
The next sales review should spend less time rebuilding the report and more time deciding what to do with it. Which customers need a conversation? Which deals have a credible next step? What is getting in the team's way?
When the records can support those answers, the pipeline starts earning the attention you give it.
If your team keeps having the same argument about the numbers, Lewis Lab's performance diagnostic work starts with the records behind it. For dealer groups, that can mean examining lead handling, CRM outcomes, and differences between stores, then agreeing on what to address first.
Discuss a CRM and Pipeline Review