Choose your next analysis What could change the decision?

Consider a fictional data-centre operator that must serve a tenant by September. It can expand its own site or lease capacity elsewhere. The tenant will go to another provider if neither route is ready in time.

You could investigate market growth, power availability, operating margin, tenant demand or competitors. All are relevant to the industry. Only some are immediately useful for this decision.

First ask what you already know. If the tenant has signed a volume commitment, spending another analysis estimating its demand may add little. If the power-connection date is unknown, it could determine whether the own-site option exists at all.

Suppose the site expansion earns £2 million more annual contribution than leasing. That is attractive. But if power cannot arrive until December, it misses the September commitment.

The immediate next question is therefore:

“What is the earliest feasible date for power and commissioning? If it is after September, I would move to leasing or a temporary bridge before comparing long-term returns.”

You have explained the dependency. Detailed margin analysis belongs after the route passes the delivery test, or as part of deciding whether a staged solution is worthwhile.

A constraint does not need to be financial to deserve priority. A permissions restriction, minimum service standard or unavailable distribution channel can play the same role.

The biggest cost line is not automatically the best first analysis. A £20 million input cost fixed by contract this year may offer less immediate opportunity than a £3 million avoidable loss.

Likewise, a highly uncertain market forecast may deserve less attention than a nearly settled implementation cost if that cost sits just above the client’s investment limit.

Ask what result would make you choose differently. Start with a consequential question you can resolve with the available evidence.

If you cannot answer, the analysis may be interesting but premature.

Imagine a different investment with annual net benefits estimated between £300,000 and £400,000. The client requires at least £250,000.

The whole benefit range clears the hurdle. If the estimate includes all material costs, refining the benefit to the nearest £1,000 is unlikely to change the decision. A remaining implementation constraint may deserve your time instead.

Now change the range to £200,000–£300,000. It straddles the hurdle. The benefit estimate has become decision-critical, so investigate the assumption responsible for that range.

You can make this explicit:

“The decision changes at £250,000. I would check the utilisation assumption first because it moves annual benefit across that threshold.”

This is more useful than saying “I would get more accurate data.” It identifies the precision you actually need.

For the data-centre case, a working plan might look like this:

There is no need to manufacture numerical scores for “impact” and “uncertainty.” The dependencies provide a clearer explanation than a weighted matrix with invented weights.

Suppose the interviewer says power is available in July and commissioning takes one month. The timing test passes. Move to the economics of the feasible routes.

If power arrives in October, examine whether leasing can bridge the gap and what that does to the investment. If neither option can serve September, consider renegotiating the commitment or declining it.

Do not keep following the original branch order simply because you announced it. A useful work plan evolves as you learn.

In an interviewer-directed case, answer the question you are given. You can still show priority within that task by choosing the most relevant comparison and explaining its implication. Leadership does not require arguing for your preferred sequence.

If the interviewer cannot provide the connection date, give a conditional path:

“If commissioning can finish by September, I would compare own-site returns with leasing. If it cannot, leasing becomes the immediate route. Can we assess its availability and cost while keeping the timing condition open?”

You preserve the unresolved issue and continue with work that has value in either outcome. Request the smallest comparison that could change your next action, rather than asking for every available dataset.

Take any case prompt and list three possible next analyses. For each, finish: “If the result is ___, I would ___.” Then choose the one with the clearest consequence.

Try Fintech Support Costs After a Card Launch. Commit to a first investigation and explain what its answer could change. After the result, state what you now believe and why your next investigation follows.