Modelling test

The format is an information pack, a spreadsheet and somewhere between one and three hours, usually with a short set of questions on the end. It is most common in private equity and experienced hire processes, appears in some graduate investment banking assessment centres and not others, and is worth asking about rather than assuming.

What gets marked is narrower than candidates expect. Whether a cell is a live formula or a typed result. Whether that formula references the cells it should, so that changing an assumption actually moves it. Whether the answer sits inside a defensible band, since any question containing a judgement call has a range of right answers rather than a single one. And the written judgement, which is where the reasoning gets credited.

The consequence is that a smaller model which runs from assumptions through to an answer beats a larger one that stops halfway, every time. The unfinished build cannot be marked at all, because the output being tested was never produced, however good the part that exists.

Simplify visibly rather than invisibly. One blended debt tranche instead of three, interest on the opening balance to avoid the circularity, working capital as a percentage of revenue rather than by day count, each written into the assumptions block. Stated, a simplification reads as a decision. Unstated, it reads as something you did not know about.

The reliable failure is spending the first stretch perfecting the revenue build, because that is the part that feels like analysis, and never reaching the output the exercise was set to produce.