Back to Blog
Jaydeep Sharma
Jaydeep Sharma

Co-Founder & Director

Published Aug 17, 2026 · 5 min read

Designing a Claim Reference Numbering Scheme That Doesn't Break

A claim reference number looks like a small detail until it's wrong. Here's how to design one that holds up as a firm scales across offices, insurers and years.

Insurance
Claim reference numbering scheme design

A claim reference number looks like a small detail until it's wrong — a duplicate, a collision across offices, or a manually typed number nobody can search for. Here's how to design one that holds up as a firm scales across offices, insurers and years.

Why Manual Numbering Fails

The obvious first approach is a running number, assigned by whoever opens the case. It works until there are two offices opening cases at the same time, or a financial year ends and nobody agrees whether the sequence resets, or someone types a digit wrong and two claims end up sharing a reference. None of these are edge cases — they're the normal failure modes of manual numbering at any real volume.

What a Good Scheme Encodes

A reference number should tell you something true about the claim without anyone having to look it up, and it should be impossible to duplicate by construction rather than by discipline. That means composing it from facts that don't change about the case, plus a sequence:

  • Insurance company — which insurer the claim belongs to
  • Claim type — fire, motor, engineering, or whichever category applies
  • Office — which branch or location opened the case
  • Loss location — where the loss actually occurred, which isn't always the same as the office
  • Financial year — since most claims operations report on a financial-year basis
  • Sequence — a number that increments within the combination of everything above

Generating the Sequence Without Collisions

The sequence has to be generated by the system, not typed by a person, and it has to be safe against two cases being created at the same moment in different offices. The reliable way to do this is to scope the sequence to the exact combination of company, claim type, office and financial year — so each combination has its own counter — and generate the next number inside a database transaction that locks that specific counter until it's committed. Two simultaneous requests for the same combination then queue rather than collide.

It's worth resisting the temptation to use a single global counter for simplicity. A global counter is easier to build and immediately becomes a bottleneck and a single point of contention the moment more than one office is creating cases at once.

Handling the Financial Year Rollover

Because the financial year is part of the reference, the sequence naturally resets when the year changes — a new financial year is a new counter, not a continuation of the old one. The one thing worth deciding explicitly, rather than by accident, is what happens to a claim that's still open when the year rolls over: its reference should stay exactly as it was assigned, since a reference identifies when a claim was opened, not its current status.

What Happens When a Claim Moves

Claims occasionally need to move office, or a claim type gets reclassified partway through. The reference number should not change when this happens — it was assigned at intake and identifies the claim's origin, not its current state. Any reassignment should be recorded as an event in the audit trail, not as a rewrite of the reference itself. Changing a reference after the fact is how systems end up with two different documents pointing at two different numbers for what is actually one claim.

Case in Point

This system generates a six-part reference automatically for every case — company, claim type, office, loss location, financial year and sequence — removing manual numbering errors entirely across an insurance surveyor firm's full operation.

Read the full case study.

Frequently Asked Questions

What should a claim reference number include?

Enough components to make it both unique and readable at a glance — commonly the insurer, claim type, office, financial year and a sequence number. The right components depend on how the organisation actually needs to search and report on claims.

How do you prevent duplicate reference numbers when multiple offices create claims at once?

Scope the sequence to the exact combination of fields that make a claim unique — company, claim type, office and financial year, for example — and generate each number inside a database transaction that locks that specific counter, rather than using one global counter for everything.

Should a claim's reference number change if it moves office or is reclassified?

No. The reference identifies when and where a claim originated. Changes to a claim's status, office or classification should be recorded as audit trail events, not by altering the original reference.

Insurance
Jaydeep Sharma

Jaydeep Sharma

Co-Founder & Director

Jaydeep Sharma is Co-Founder and Director at Nullpreneurs LLP and works with Stacknyu’s team to shape practical software products, delivery operations and long-term client partnerships. His focus is on turning business requirements into useful, maintainable digital systems.