Insights CRM and customer data
Why your customer reports disagree
Different customer totals can come from different definitions, even when each report is calculated correctly. Reliable reporting starts with agreement on what is being counted, which records support it and who owns changes to the definition.
Imagine a property business preparing its monthly report. Marketing counts people who have made an enquiry. Sales counts purchasers with a signed contract. The service team counts people with an active support relationship.
Each group has a useful number. Labelling all three as customers creates confusion when those numbers meet in a management report. Moving them into one dashboard will make the disagreement easier to see, but someone still needs to resolve its meaning.
Define the decision behind the number
Start with the business question. A sales leader assessing follow-up needs different information from a finance team assessing outstanding payments.
Give distinct concepts clear names. Enquiries, qualified opportunities, contracted purchasers and active service accounts should not become interchangeable simply because they relate to the same person.
For each important measure, document the event that includes a record, the event that removes it and the time period used. Establish how cancellations, duplicates and later corrections are treated.
The definition should be short enough to use in a working discussion. If people still cannot apply it consistently to a real record, it needs more work.
Separate identity from business status
Recognising the same person across systems is a different problem from agreeing what that person's relationship with the business means.
A customer data platform can help combine records into a unified profile. That requires matching rules and choices about which source fields to retain. Microsoft's documentation describes these as explicit steps in data unification.
Even with a unified profile, a person may hold several roles. They can be a prospect for one service and an existing customer for another. A shared email address may also belong to several people, so matching records by email alone can be inappropriate.
The profile needs to preserve those distinctions and represent the underlying relationships accurately enough for the work it will support.
Assign a source and an owner
One application does not have to own every fact. The CRM may own opportunity status while a finance platform owns invoice balances.
Agree the authoritative source for each important field or measure. Then define how updates reach the reports and other systems that use them. A report should make its refresh time visible when the delay affects a decision.
Nominate a business owner who can settle questions about the definition. Technical teams can implement a rule, but they need an authorised answer when sales and finance use the same term differently.
Keep changes visible. If a definition changes halfway through a year, historical comparisons need to explain whether earlier figures have been recalculated.
Test the rules against actual work
Review a small set of representative records together. Include a cancelled transaction, a duplicate contact, a customer with several accounts and a record changed after the reporting period closed.
Ask each team to apply the proposed definition. Differences reveal where the rule or source data remains unclear. Resolve those questions before distributing a new report more widely.
The same discipline matters when customer data is used for automation or AI. A system selecting people for follow-up needs an agreed definition of eligibility and appropriate permission to use the information. Combining records does not establish that permission.
Once the definitions are stable, the technology requirement becomes clearer. The business may need better CRM configuration, connected customer data, reporting changes or a combination of these. The choice can then follow the work the teams need to do with the information.
Discuss your customer data requirements
Further reading: Microsoft on customer data unification.