How a tenfold price error was stopped before it was written: the nine-gate check

Situation

A panel value of 381.5 was seen beside a value of 3,815 prepared for writing. The gap was a tenfold risk created by reading a comma and decimal separator differently. If the on-screen number had been accepted alone, the wrong value could have reached the writing stage. The misleading figure was a panel value that looked ordinary when not read in its proper format.

This observation does not describe a customer brand or commercial result. It identifies the need for evidence at the last data-entry step as well. Even when a value appears copied from a source, the question of whether its written form matches the recorded form must be asked plainly.

Measurement

The check occurred in a fifteen-minute round using an independent record. The panel value and value to be written were compared in separate fields. When they did not match, the record carried the warning “panel value does not match the record.” The warning stopped writing before anyone tried to explain the change.

The independent record made the review repeatable. The next round could follow the same comparison in the same order, so the difference did not depend on one person noticing it. Measurement has no authority to correct a number. It shows the threshold at which a mismatch must go to human review.

In bulk price updating, decimal and thousands separators are not merely visual formatting; they alter the meaning of the value prepared for writing. When the source panel and the writing field use different conventions, the comparison identifies how each value is being read. A format conversion is never assumed silently, and the panel does not replace the independent record. When source and writing value disagree, the review does not guess the correction. It exposes the mismatch, refuses the write, and sends the evidence to the authorised person, keeping a plausible-looking number from bypassing source verification.

Finding

In the cosmetics · 67 products · 2026 record, the decimal error 381.5→3,815 was rejected. The reason was a mismatch between panel and independent record. No sales, revenue, inventory, or financial effect can be inferred from this. The record says only that the pre-writing check worked and the mismatched value was not published.

The finding showed that speed cannot replace verification. Even a value that appears right is matched across sources. When there is no match, writing stops rather than producing an explanation or continuing from a guess. This discipline does not promise the same result for other products; the outcome depends on the product and market.

Rule

The 2026 rule is this: a price to be written is not sent until it matches the independent record through the nine-gate review. The gates were retained as a workflow that checks the input value, its format, and its recorded counterpart separately. The rule text stayed beside the round record so it could be accessed by a reviewer, not only remembered by the person who knew it.

The rule does not write a number by itself or promise a commercial outcome. When a mismatch is found, the flow stops, the record is reviewed, and an authorised person decides. This creates a clear safety boundary between the need for speed and a difficult-to-reverse written value.

The nine gates form an accountability chain from source to writing rather than a list that repeats one question. The source, display convention, and recorded counterpart must be understood before the next step. A mismatch message gives the reason for refusal; it does not propose a corrected value. Replacing one uncertain format with another guess would not be a review. The flow can be considered again only after a person returns to the source and confirms a match. Without that confirmation, the held state remains visible and the next round can understand exactly why writing stopped.