Problem guide
The same customer exists three times in your systems
Duplicate customer records happen when every system that touches a customer lets someone create them, and each person types the name a little differently. Decide which system owns the customer record, stop creating customers anywhere else, and clean up the worst duplicates by hand while you plan a lasting fix.
What it usually looks like
A property management company has twelve buildings under contract. In the scheduling system it is listed once by its legal name, twice by the names of two buildings, and once under the property manager's personal name from an old emergency call. Billing has it under an abbreviation. When the company pays a batch of invoices with one check, the office applies it to the first record it finds. The other records show as overdue, and the collections clerk calls the property manager about bills she already paid. Nobody can say what the account is worth in a year.
Signs you have this problem
- A customer search returns the same company under several spellings or addresses
- Job history and invoices for one customer are split across records
- A payment gets applied to the wrong record and the real account looks overdue
- Nobody can give a straight answer to how much work a customer did with you last year
- Each system has its own customer number for the same company
What usually causes it
- Customers are typed separately into the quoting, scheduling and billing systemsThe re-entry trap
- Whoever takes the first call creates a new record rather than searching, because searching is slow or unreliableThe handoff gap
- Sites, billing addresses and parent companies are handled as special cases with no ruleThe exception swamp
What to try this week
- Measure the duplicates in one systemExport the customer list from your billing system and sort by name, phone and address. Count how many customers appear more than once. Do the same for one other system.
- Name the owning systemPick the one system where a customer is created first and for real. Everyone else copies from it, and nobody creates a customer anywhere else.
- Write a naming ruleSet a short rule for how names, sites and billing addresses are entered, and how a parent company relates to its locations. Put it where the people creating records can see it.
- Search before you createAs a temporary habit, require a search by phone number and street address before any new customer is added. Most duplicates start with a skipped search.
- Merge the worst ones firstStart with the customers you bill most often. Merging the busiest duplicates fixes the most reports and the most payment mistakes for the effort.
Where do-it-yourself stops
These steps slow down new duplicates and clean up the ones that hurt most. They do not stop each system from keeping its own copy of the customer, so the copies drift apart again every time someone updates one of them. A lasting fix is a tool built around your own quoting, scheduling and billing systems, so a customer is created once and every system reads the same record. That is what a Workflow Spec designs.
Questions people ask
Why does the same customer appear several times?
Because more than one system lets people create customers, and each person enters the name, site or address slightly differently. Nothing checks for a match before the record is saved.
Is deduplication a one-time cleanup?
The cleanup is. The prevention is not. Without a rule about where customers are created, duplicates come back within months.
How should we handle a company with many sites?
Decide whether the billing customer and the service site are separate records, and apply that rule every time. Mixing the two approaches is a common source of duplicates.
What do duplicates actually cost?
Misapplied payments, collection calls to customers who paid, and reports that undercount your best accounts. Count your duplicates and trace a few payment errors to see your own cost.
Can we just merge records in the software?
Most systems can merge within themselves. The harder part is the other systems that still hold their own copy, which a merge in one place does not fix.