This is untimed educational practice for one applied skill. It is not a clinical, diagnostic, employment, or professionally recognized assessment, it does not produce a score or credential, and for the judgement and safety families it is not a substitute for your employer's own policies, training, or legal duties.
Checking first turns a suspicion into a fact, and telling the owner leaves the decision with the person accountable for the report while there is still time to act. Quietly correcting it is the tempting alternative because it looks decisive and fixes the number, but it removes the owner's chance to catch an error in your reasoning, and nobody afterwards can tell which figures were changed or why.
Private, specific, and forward-looking gives the person a realistic chance to change something they may not have noticed. Raising it in the meeting is the strong alternative (it is direct, and the behaviour did happen in public), but an audience turns a correctable habit into a status contest, and the most likely response is defensiveness rather than a different meeting next week.
This makes real progress on the certain part while converting the vague unease into questions someone can answer in a sentence. Waiting is the plausible safe option and the expensive one: it spends two of the five days you were trusted with, and the questions would still be unanswered on the day they return.
Told on Monday, the slip is still a choice: someone can add help, cut scope, or reset the client's expectation. Absorbing it privately is the tempting answer because it shows ownership and effort, but it converts a manageable problem into an unavoidable one, and the client hears about it later rather than sooner.
Complying keeps the team's output consistent while you gather the one thing that can actually change a decision: a real case. Carrying on with the old process is the tempting option because you may well be right, but two processes running silently in parallel make the team's output inconsistent, and the owner never receives the evidence that would have improved the new one.
One short conversation usually surfaces a blocker somebody else can clear, and it keeps the decision to escalate a joint one. Going to the manager first is the strong alternative because the deadline is real, but it skips the fastest available fix, and it teaches the colleague that a problem with you arrives via their manager.
Committing to a decision you argued against, with an agreed trigger for review, keeps the team moving and keeps your dissent useful. Continuing to argue is the tempting response, because conviction feels like integrity, but the objection is already on the record, so repeating it stalls the work and spends credibility you will want for the next disagreement.
You keep your own account accurate, you help with the thing that actually matters, and you leave them the chance to self-report, which is nearly always treated better than being found out. Going straight to their manager is the option that feels most rule-abiding, but it removes that chance without protecting anyone further, and it makes the next colleague less likely to tell you anything.
A bounded amount of your time unblocks them and leaves them able to do it alone next time, which is the only version of help that scales. Doing it for them is the generous-looking alternative: it fixes today, leaves the gap in place, and hides from their manager that support was needed at all.
Afterwards and in private you can be precise, and the record, which is what counts at review time, actually gets corrected. Correcting it in the room is the tempting response because the error was public, but it turns a probable oversight into a visible dispute and still leaves the written record untouched.
The client is currently making decisions on a wrong conclusion, so speed matters more than comfort, and arriving with a proposed correction makes it a five-minute decision rather than a crisis. Fixing the internal copy is the plausible half-measure, it feels like putting things right, but the only copy that matters is the one the client is reading.
Declining protects your judgement; disclosing protects the procurement, because the organisation can then see whether this supplier makes such offers routinely. Declining quietly is the near-miss and is the one most people choose: you did the right thing personally, but the pattern stays invisible, and if the offer surfaces later your silence is harder to explain than the offer itself.
Naming the specific policy often resolves it immediately, because most such requests are a misreading rather than an instruction to break a rule; asking for it in writing changes the incentive if it is not. Deferring to seniority is the intuitive answer, but accountability does not transfer upward simply because the instruction came from above, the person who acts is still one of the people who acted.
A near-miss is a free warning, and it only works if it reaches the people who can change the equipment, the layout, or the procedure. A private note is the well-intentioned alternative, you have recorded it, but it is invisible to everyone with the authority to act, so the hazard is still there on the next shift.
Asking what would change their mind either produces the missing evidence or tells you the disagreement is about values rather than facts, and it keeps a legitimate route open if the risk is severe. Going over their head is the option that feels proportionate when you believe harm is coming, but the same unchanged argument usually gets the same answer one level up, and you have spent the standing you would need if the risk does escalate.
The value of the warning falls with every day it is delayed, and offering a partial delivery and a revised date lets them replan rather than simply absorb the news. Leading with the explanation is the natural instinct, but a detailed account of why it happened does not tell the other person what they now have to work with.
A status report is an input to somebody else's decisions about people and sequencing, so it has to describe the position rather than the forecast. 'On track' is the tempting answer precisely because it may turn out to be true, but it hides the only fact that would let a manager move help now, while moving help is still cheap.
Most pointless-looking steps are controls for a failure you have not seen yet, so finding out what it guards against is what turns a complaint into a proposal somebody can approve. Skipping it to test the theory is the tempting empiricism, but the experiment only fails when the rare event it prevents finally happens, and by then it is the incident rather than the experiment.
Splitting the feedback lets you improve on the accurate half immediately and turns the disputed half into something checkable rather than a matter of opinion. Accepting all of it is the option that feels gracious and low-friction, but it leaves an inaccurate characterisation standing in the record and gives you nothing concrete to change.
A measured number turns 'this is annoying' into a decision with a payback period, which is the form a manager can approve or decline on its merits. Building it quietly is the initiative-flavoured alternative, and it is how organisations acquire undocumented tools that only one person can run, audit, or fix when they are on leave.