The Certificate That Stopped Needing Reprints

Background

Every piece is accompanied by a printed certificate. Over an eighteen-month period, reissue requests averaged 16.4 per cent of certificates printed — roughly one in six — and each reissue costs materials, handling, and a delay measured in days.

The reissue rate had been treated as a print quality problem for over a year, and print quality had been improved twice with no effect on the rate.

Problem Statement

Identify the actual causes of certificate reissue requests and reduce the reissue rate, without reducing what the certificate asserts or making it easier to forge.

Objectives

1. Establish the true distribution of reissue causes, rather than the assumed one.
2. Reduce reissue rate below 5 per cent within one quarter.
3. No reduction in the information a certificate carries.
4. No increase in production cost per certificate.

Constraints

• Certificates already issued could not be recalled or invalidated.
• The physical format was fixed by the enclosure it ships in.
• Any change had to be legible to someone holding a certificate from before the change, without explanation.

Approach

The first step was to stop guessing. For six weeks, every reissue request was logged with a free-text reason, taken from the requester in their own words rather than selected from a list. Categories were derived afterwards from the text, not imposed on it in advance.

This was the whole of the method, and it was sufficient.

Implementation

Logging ran for six weeks and captured 71 requests. Categorisation was done twice, independently, and the two passes were compared; agreement was 66 of 71, and the five disputed items were resolved by re-reading the original text.

The change that followed was made in a single release: the ownership line was reworded from an assertion about the current owner to an assertion about the piece and its first registration, with current ownership moved to the digital record only.

Metrics

Results

Reissue rate fell from 16.4 per cent to 1.8 per cent over the quarter following the change, against an objective of 5 per cent.

The result is almost entirely attributable to a single finding: 79 per cent of reissue requests were caused by the certificate asserting something that changes — who owns the piece — on a medium that cannot be changed. Print quality, which had received two rounds of investment, accounted for 6 per cent.

What Worked

Free-text reason capture. Had reasons been collected from a predefined list, the list would have been built from the existing assumption and would have contained options for paper weight and ink density. The dominant cause would not have had a box to be recorded in.

Moving the mutable fact to the mutable medium. This is the entire fix, and it made the certificate assert less while making it more durable — the certificate now says only things that were true at printing and will remain true.

What Did Not Work

The six-week logging period was longer than necessary. The distribution was clear at three weeks and did not shift afterwards; three weeks of delay were spent confirming something already visible.

Damage in transit, at 12 per cent, was left entirely unaddressed and remains the largest untreated cause. It was outside the stated scope, but scoping it out was a choice and it is now the ceiling on further improvement.

Future Improvements

Address transit damage, now the dominant remaining cause. Preliminary observation suggests it is concentrated in a single enclosure configuration, but this has not been measured and should not be acted on until it is.

Apply the same free-text method to two other processes where the cause of failure is currently assumed rather than recorded.

Re-examine the remaining assertions on the certificate against the same test: does this statement have any chance of becoming false while the certificate still exists? Ownership was the obvious case. It may not be the only one.