Case study · United Kingdom · white-label for B4B
No more duplicate customers in Sage 200
At Collins Nets, a UK fishing nets and supplies business, duplicate customer records were appearing faster than the team could clean them up. We built a system that detects, scores and merges them - white-label development for the B4B agency.
Client and context
Collins Nets takes orders through multiple channels, and every channel and every manual edit is a chance to duplicate a customer in Sage 200. Over time the database filled with records of the same person or company - with different spellings, addresses or phone numbers.
Challenge
Manual clean-up stood no chance: new duplicates appeared faster than old ones could be resolved. What was needed was an automated system that matches records confidently but leaves ambiguous cases to a human - and that does not damage data in the company's accounting and sales system.
Solution
We built a duplicate merge system connected to Sage 200. Every night it syncs customer records from Sage into the CRM, and a weighted algorithm scores matches across five signals: phone number, postcode with surname, address, company name and email domain. A review panel shows the pairs with colour-coded confidence levels, and a person makes the decision. Phase one ran read-only; phase two added automatic deactivation of records in Sage, API-driven merging and an undo function.
Scope of work
Who was responsible for what
A white-label delivery: the client worked with the B4B agency, and we were responsible for development.
What we built
- Nightly synchronisation of customer records from Sage 200 into the CRM
- A weighted matching algorithm across five signals
- A review panel with confidence levels and a human decision
- Phase 2: deactivation of records in Sage, API-driven merging, undo
On the partner's side
- Client relationship and project management
- Agreeing the phases: read-only first, then write
Deployment process
Read first, write later
Two phases, so the system could first prove the accuracy of its matches and only then touch the data in Sage.
Read-only
Sync with Sage, the algorithm and the review panel - with no changes to source data, to build trust in the matches.
Writing to Sage
Automatic deactivation of duplicates, record merging through the API and the ability to undo an operation.
Deployment stack
Let's talk
Data drifting apart between systems?
Tell us where duplicates and errors arise - we will show you how to catch them automatically. 30 minutes, no obligation.