Which address is right?
The CRM says the customer is at the old unit on the trading estate. The accounts system has the new office. The delivery system has a third address with a typo in the postcode. The customer is called "Smith and Sons" in one place, "Smith & Sons Ltd" in another and "SMITH SONS" in a third.
Invoices go to someone who left two years ago. A delivery goes to the old unit. Sales call a contact who no longer works there. Each time, someone fixes the record in the system in front of them, and the others stay wrong.
Why the records drift apart
Each system was filled in separately, by different people, at different times. Nothing links the record for Smith and Sons in one system to the record in another, except the name, and names are unreliable.
Even where an integration exists, it may copy records once at creation and never again, or match on name and create a new record every time spelling changes. Duplicates within a single system make it worse: the CRM may hold three versions of the same company, each with part of the history.
The root cause is almost always the same. There is no shared identifier and no agreed master for each piece of information.
Staff habits add to it. When a system will not accept a change easily, people put the new address in a notes field, or create a fresh record rather than edit the old one. Each shortcut is sensible in the moment and leaves one more version of the customer behind.
What mismatched records cost
- Invoices and statements sent to the wrong person or address, and paid late as a result
- Deliveries to old addresses, with the cost of re-sending
- Sales and service staff working from different versions of the truth
- Reports that count the same customer several times
- Data protection risk from holding out-of-date personal details
- Time spent checking which system is right before every call or letter
The damage to trust is harder to see. A customer who has told you their new address twice and still gets post at the old one draws their own conclusions about how organised you are.
How we bring them into line
- Extract. We pull the customer records from each system into a working area, with their IDs and last-updated dates.
- Match. We match records across and within systems using company numbers, VAT numbers, email domains, phone numbers, postcodes and cleaned-up names, scoring how likely each pair is to be the same customer.
- Review. Confident matches are accepted. Uncertain ones go to a simple screen where your team confirms or rejects them. Nothing merges on a guess.
- Choose the master per field. We agree, for example, that the accounts system owns the billing address and the CRM owns contacts. The best current value is picked using those rules and the last-updated dates.
- Write back and link. Each system is updated with the agreed values and stores the IDs of the matching records elsewhere, so future syncs match on ID, not name.
- Keep it in step. An ongoing sync pushes changes from the master to the others, and new records are checked for likely duplicates at creation.
| Field | Typical master | Why |
|---|---|---|
| Legal name and company number | Accounts | Needed for invoicing and credit checks |
| Billing address | Accounts | Where invoices must go |
| Delivery addresses | Order or delivery system | Updated by the people who deliver |
| Contacts and roles | CRM | Maintained by sales and account managers |
| Marketing preferences | Email platform or CRM | Where consent is captured |
What your team sees afterwards
Look up a customer in any system and the details agree. Change an address in the master and it reaches the other systems without anyone copying it. New duplicates are caught when they are created, not discovered at year end.
Reports count each customer once. Letters go to the right person at the right address.
The review screen stays in use at a low level, catching the occasional near-duplicate when someone creates a customer that already exists. That small, ongoing check is what stops the records drifting apart again over the following years.
Signs your customer records need this
- The same customer has different details in different systems
- Invoices or deliveries go to old addresses
- The CRM holds several versions of the same company
- Staff check two systems before contacting a customer
- Customer counts differ depending on which report you look at