Quality Review in Professional Services: Who Checks the Work
Most review in professional services checks whether a document reads well. Very little of it checks whether the conclusion is right.
Ask a firm how it assures quality and you will usually be shown a workflow: the manager reviews, then the director, then the partner signs. Ask what each of those people is looking for and the answers overlap almost completely. The result is three passes that all check the same things, mostly presentation, and none that checks the thing most likely to be wrong.
Review is expensive. It consumes the most costly hours in the firm and it delays every deliverable. It is worth designing so that each pass looks for something the previous one could not.
What each level is actually for
- The preparer's own check: has every figure been traced to its source, does every statement have support in the file, and does the document say what the analysis found rather than what was expected.
- The detailed review, usually a manager: is the method sound, is the evidence sufficient, and are the exceptions and inconsistencies explained rather than smoothed over. This is the only level that can realistically look at the working papers.
- The engagement leader's review: is the conclusion supportable, is the risk correctly characterised, and would the firm be comfortable defending this in front of a regulator, a board or an opposing expert.
- The independent or second-partner review, where required: has the engagement leader's judgment been anchored by proximity to the client, and would someone with no relationship reach the same conclusion.
When a second partner is genuinely needed
A second-partner review is the most expensive control a firm has and it should be reserved for engagements where the exposure justifies it. Useful triggers are that a third party will rely on the output, the client is listed or regulated, the conclusion is finely balanced, the fee is materially above the firm's norm, or the engagement leader has served this client for a long period.
That last trigger is the least popular and the best supported. Long relationships produce genuine understanding and genuine blind spots at the same time, and the person best placed to see the blind spot is someone without the relationship.
Making review find defects
Three practices lift the yield of review substantially and none of them costs more time.
First, review against the acceptance criteria rather than against taste. A reviewer given the criteria checks whether they are met; a reviewer given only the document rewrites the introduction. Second, ask the reviewer to state what would have to be true for the conclusion to be wrong, which converts a reading exercise into a test. Third, require that the most significant judgment in the deliverable is identified by the preparer, so the reviewer's attention lands where the risk actually is rather than where the prose is weakest.
Recording review so it counts
A review that leaves no record did not happen, as far as anyone examining the file afterwards is concerned. The record should show who reviewed, what they reviewed, when, and how each point raised was resolved. Points closed with "discussed and agreed" are the ones that look worst in hindsight, because they document a conversation rather than a resolution.
Equally important is that review points are resolved before the deliverable is issued rather than after. A file containing open review points and an issued report is the clearest possible evidence that the control did not operate, and it is a surprisingly common finding.
The cost nobody measures
Review rounds consume margin invisibly, because the time appears as ordinary chargeable effort rather than as rework. A firm that tracks review rounds per deliverable, and the reason for any round beyond the second, usually discovers a small number of recurring causes: an unclear brief, a template that does not fit the work, or a reviewer who edits rather than reviews.
All three are fixable, and fixing them returns more than any exhortation to be more careful. The measure is worth adding for that reason alone: it turns a quality conversation into an operational one.