A delivery promise can change for a legitimate operating reason. The customer may need a new arrival expectation, and colleagues need current information to organise the next action. I do not think a historical performance account should prevent that update. I do think the update should remain distinguishable from the undertaking that preceded it. When the revised date replaces the original everywhere, the record can make service appear better without showing that any journey actually became faster.
The fictional 2023 review provides a simple example. Of twenty thousand shipments, sixteen thousand met their original promises and four thousand did not. Three thousand of those originally late shipments had their promised dates revised after departure and qualified as on time against the updated dates. The original-promise result was 80%; the updated-promise result was 95%. Both percentages can be calculated correctly from their stated references. The difficulty begins when one is presented without explaining which question it answers.
Two promises can answer two legitimate questions
An original commitment helps describe performance against what was first undertaken. A current promise can help service colleagues discuss what is now expected. Those purposes need not compete for the right to exist. The account becomes misleading when the current view acquires the meaning of the historical one without retaining its reference point. I would not solve that problem by forbidding revisions, because an outdated operational expectation can also be unhelpful. I would preserve the distinction and make the purpose of each view explicit.
Mara Ellervane challenged the revised-only presentation in the case. The service sponsor wanted the measure to reflect the promise colleagues were currently discussing with customers. I understood that preference, but it could not justify describing the difference as a delivery improvement. The shipments and their arrival events had not changed. The reference dates had. Keeping both accounts cost us a single uncomplicated headline, while allowing the current service conversation and the earlier undertaking to remain available for different decisions.
The data function should also avoid making a moral or legal conclusion from the existence of a revision alone. A change may have an explanation that matters to its interpretation. Commercial and service owners need to establish what happened and what was agreed through their responsible processes. My role is to retain the relevant versions, timing and recorded reason so that the evidence can support that assessment. A date history is not a substitute for the contractual context or authority required to decide its consequences.
Preserve the revision as an event rather than a replacement
The design choice I favour treats a revision as something that happened, with a relationship to the promise before it. A current view can select the latest appropriate state for action, while the history remains available for evaluation. This requires more than retaining an old value in an inaccessible file. The account should make the difference understandable to the people who use the measure, including when a short management table is separated from the system in which the history is stored.
US$2.8 million supported the fictional work to retain original and revised promises, change times and recorded reasons. The output was an evidence account, not a service-recovery result. I would describe it that way in a leadership profile. A data programme can help operating colleagues make better-informed decisions without claiming the outcomes of interventions it did not measure. The percentage difference itself cannot be converted into saved time, retained revenue or satisfied customers merely because it is large and easy to communicate.
There are practical questions about which original promise the account means. A business may have an initial request, an accepted commitment and later updates. The definition must be established for the purpose of the measure rather than assumed from a field named original. I would ask the responsible teams to identify the event at which the relevant undertaking was made and the source supporting it. The fictional case states that reference explicitly; another institution would need to establish its own evidence and meaning.
A series also needs to explain changes in its reporting basis. If an earlier month uses first commitments and a later month uses current promises, an apparent improvement may partly reflect the measure’s construction. I would retain a clear account of that change before inviting a comparison. This is not a reason to preserve an unsuitable measure indefinitely. It is a reason to distinguish an improved definition from an improved service, and to avoid presenting their effects as though they were the same development.
Explain the performance claim before celebrating the percentage
For a senior reader, the essential question is what the percentage asks them to believe. Does it describe performance against an initial undertaking, the current position of the shipment or something else? The answer should be available in ordinary language close to the figure. A concise definition is often more valuable than a technically elaborate appendix that leaves the headline unchanged. The data leader has to understand the decision well enough to identify the distinction that materially affects interpretation.
I would expect operating colleagues to challenge the practicality of the account. Maintaining history has a cost, and not every revision detail will be equally useful to every recipient. The response should be proportionate: retain the evidence needed for the relevant questions and present it in a usable way. The choice is not between an overloaded report and a misleading single number. Different views can share a coherent underlying history while exposing only the context each decision actually needs.
The board should not have to reconstruct a promise ledger to understand a material service claim. Management can explain the measure, important changes and remaining limits through its established reporting arrangements. A director with relevant experience can ask the clarifying question without becoming the operational owner of the data. That distinction matters to me because a strong technical perspective should support collective oversight, not justify creating a parallel management process outside the company’s arrangements.
I would also plan the explanation for people who saw the earlier, simpler account. Changing the measure without discussing why it changed can create another misunderstanding, such as an apparent sudden decline in service. The correction should identify which part of the picture was redefined and what the underlying delivery evidence still says. A trustworthy account does not hide the history of its own interpretation merely because the revised presentation is now more accurate.
My position is that original and current promises should remain connected rather than compete to overwrite one another. The enterprise needs current information to act and a faithful history to understand what occurred. The twenty-thousand-shipment example shows why a higher on-time percentage can be a consequence of changed reference dates rather than changed journeys. Preserving that distinction gives service recovery, performance review and leadership communication a shared account whose meaning can survive the pressure for one attractive number.