The Adviser's Audit Trail: Requiring the Reasoning Behind the Recommendation

The conclusion is not the advice

A board paper arrives with a clear recommendation: proceed, defer, restructure, sell. The reasoning is compressed into a handful of bullet points and a slide that implies rather than demonstrates. The decision-makers around the table are left to accept the conclusion or reopen the entire engagement. Neither is satisfactory. The first weakens governance; the second wastes the specialist input that was bought precisely to inform the decision.

The remedy is not a longer report. It is a clearer audit trail. Specialist advisers should be asked, as a condition of the engagement, to show the evidential basis on which a material recommendation rests. That means distinguishing what has been verified, what has been assumed and modelled, and what reflects professional judgment. The distinction matters because each category carries a different burden of proof and a different route to challenge.

Why boards need the working, not just the answer

Specialist advice is valuable because it brings expertise the board does not hold in-house. But expertise is not self-authenticating. A recommendation may be sound, yet the board cannot discharge its duty by treating it as opaque. Under the UK Corporate Governance Code, directors are expected to exercise independent judgment and to satisfy themselves that the information they receive is adequate. A conclusion-only paper makes that difficult.

The evidence pack is also a discipline for the adviser. It forces the engagement team to separate evidence from inference. It exposes the points at which the analysis depends on assumptions that the board may wish to test. And it creates a record that can be revisited when circumstances change, or when a later reviewer asks why a particular course was taken.

This is not about distrust. It is about making specialist advice usable by the body that must act on it.

What an evidence pack should contain

A workable evidence pack need not be lengthy. For most material recommendations, it should contain five elements:

• A one-page statement of the recommendation and the decision it supports.
• A verified-facts schedule: the data, documents, interviews and tests that have been checked, with sources and dates.
• An assumptions register: the modelled assumptions, their basis, the range of plausible values and the effect on the recommendation if they move.
• A judgment note: the professional judgments made, the alternatives considered and the reasons for preferring the chosen course.
• A sensitivity summary: which two or three variables would most change the answer, and what the board should watch.

This structure is deliberately plain. It can sit behind the main report rather than replace it. The point is that a director should be able to trace a line from the conclusion back to the evidence and forward to the conditions under which the conclusion would cease to hold.

A simple template

For a major recommendation, the adviser should complete a short schedule with columns for claim, category, source or basis, confidence, and sensitivity. The rows would include the central claims on which the recommendation depends. A claim such as the addressable market is £X million should be marked as modelled, not verified, unless it rests on audited data. A claim that a counterparty has breached a contractual term should be marked as verified only if the relevant documents have been reviewed and cited.

The template is not a substitute for the adviser's professional narrative. It is a navigational aid. It tells the board where to look and what to press on. It also makes clear where the adviser is entitled to deference and where the board is being asked to accept an inference.

Telltale signs of unsupported advice

Certain patterns should prompt a request for the underlying working. A recommendation that relies on a single headline number without a stated basis is one. A model presented without a description of its key assumptions is another. So is a paper that uses phrases such as market practice suggests or our experience indicates without saying whose practice, which market or what experience. These phrases can be legitimate, but they are judgment, not verification, and should be labelled as such.

Other warning signs include an absence of sensitivity analysis, a recommendation that does not change even when the stated assumptions are varied, and a report that cannot identify the evidence that would cause the adviser to revise the conclusion. None of these is fatal on its own. Together they suggest that the board is being asked to rely on authority rather than reasoning.

Applying the standard proportionately

Not every engagement warrants a formal evidence pack. A small, time-bound piece of advice to a single executive may need only a short note identifying the facts relied on and the main assumption. A major transaction, a regulated change programme or a board-level strategic review should require the full schedule, with the adviser asked to present the sensitivity summary in person.

The proportionality test should be set at commissioning. The engagement letter can specify that the evidence pack is required for recommendations above a stated threshold or for any recommendation that commits the organisation to irreversible expenditure. For smaller matters, a lighter version can be used. The standard should be clear, consistent and agreed before the work begins, so that it is a condition of the mandate rather than an afterthought.

Making it stick

The evidence pack should be listed in the board papers alongside the adviser's report. The chair or audit committee should ask one question of every material recommendation: which parts of this are verified, which are modelled and which are judgment? That question is not adversarial. It is the minimum required to turn specialist advice into a governable decision.

In 2026, boards are under pressure to move quickly and to rely on specialist input that is more complex than ever. The evidence standard is a modest control that protects both sides. It gives the adviser a clear format for demonstrating rigour, and it gives the board a route to test the reasoning before it commits.