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.
Knowing what decision you are making and on what criteria is what makes a document relevant or not, and it stops you absorbing pages that cannot change the answer. Starting with the most recent document is the sensible-sounding heuristic and does nothing for you here: recency says when a document was written, not whether it is about the question you were asked.
Old evidence is still evidence about its own period, and it is often the only data available on that variable, what matters is that the bridge to today is visible and challengeable. Discarding it is the rigorous-looking move and throws away information; the discipline is labelling the assumption, not deleting the source.
A senior person's belief is a claim you have been handed alongside data that can check it, and the check is the whole point of putting both in the pack. Accepting it on authority is the tempting deference, and it converts the most testable statement in the case into an assumption nobody examines.
Saying what you excluded and why shows the reader that you read it and judged it, which is exactly what distinguishes selection from oversight. Using something from every document is the instinct 'use all the evidence' produces, and case packs routinely contain material chosen to be irrelevant, forcing a use for it dilutes the argument that does the work.
Most apparent contradictions in a case pack are definitional (headcount versus full-time equivalent, calendar year versus financial year), and resolving that gives you a figure you can defend. Averaging is the tidy-looking compromise and produces a number that neither document supports and that means nothing at all.
Utilisation has several standard denominators, and identifying which one each source used usually reconciles the two exactly. Taking the lower figure to be safe is the prudent-sounding option and is still a guess: a conservative choice between two undefined numbers is an undefined number that happens to be smaller.
A gap between what people report and what the accounts show is often the most valuable thing in the pack, and naming it is more useful than quietly resolving it. Trusting the numbers is the reflex that feels most objective, and aggregated data can easily average away a local effect that is real and expensive at one site.
Isolating the single driver turns two competing outputs into one question the decision-maker can actually answer or investigate. Presenting both is the balanced-looking option and leaves the reader exactly where you started, holding two numbers with no account of why they differ.
An interested party usually holds the best operational data about their own area, so the answer is to separate their facts from their argument. Discarding it is the sceptical-looking move and throws away the pack's most detailed source over a bias you could simply have stated.
Showing that the recommendation holds across a range converts a missing number from a blocker into a bounded uncertainty. Quietly using the midpoint is the option that keeps the analysis looking complete, and it is the most damaging: to a reader, an unlabelled assumption is indistinguishable from data.
Allocated overhead does not leave with the product, so the relevant figure is the 200,000 contribution and the question is how much of the allocation is actually avoidable. The 'never close it' answer is the strongest wrong one because it has the right instinct and states it far too absolutely: capacity, avoidable costs, and what else the space could earn all bear on the decision.
A returns 200,000 against 100,000 spent; B returns 450,000 against 250,000 spent: twice the net gain over the same horizon. Payback is the strongest distractor because it is a real criterion and A genuinely wins on it, but payback stops counting on the day the money is recovered and so ignores more than half the life of both options.
The limitation is not the sample size but the sampling: one store on one weekend afternoon selects for a particular kind of customer. 'Forty is a workable sample' is the tempting statistical answer and addresses precision when the problem is bias. A larger sample drawn the same way would be just as unrepresentative, only more confidently so.
Contribution per unit fell by 40% while volume rose by 12%, so on 100 units the total falls from 2,500 to about 1,680. 'Revenue rose' is the sharpest distractor because it is arithmetically true (revenue goes from 10,000 to about 10,080), and revenue is not what pays for anything.
A recommendation about future trading has to rest on the part of this year's profit that will happen again. Including it with a note is the transparent-looking compromise and still leaves a year-on-year comparison in which the growth is mostly a building sale.
Answer first means every subsequent paragraph has a job the reader can see, and it survives being read only halfway. Narrating the analysis in the order you carried it out is the natural-feeling structure, because it is how the work actually went, and it makes the reader carry all the evidence before learning what any of it was for.
Recommending while being honest about the margin gives the reader both your judgement and its strength, and the flip condition tells them what to watch. Presenting both without a recommendation is the scrupulously balanced option, and it hands the decision back to the person who has read less of the evidence than you have.
An assumption stated at the point it does its work can be challenged by someone who knows better, which is the only way it gets corrected before the decision. Putting it in an appendix is the technically-complete option and functions as concealment: nobody reads the appendix before deciding.
Choosing between options means comparing what each costs and how long it takes, so an answer without either is a preference rather than a recommendation. 'Implementation is separate' is the reasonable-sounding scoping argument, and it withholds precisely the information the choice turns on, even an order-of-magnitude figure changes which option wins.
Rejecting the menu is legitimate when you replace it with something: a different option, or a defined piece of work with a cost and a date. Stopping at 'inconclusive' is the honest half of the answer and leaves the decision-maker exactly as stuck as before, which is why it reads as an evasion rather than a finding.