Advisory relationships rarely fail loudly. They fade. The meetings continue, the updates are pleasant, the invoice is not large enough to trigger a review, and eighteen months later nobody can say what changed.
That outcome is preventable, and preventing it is mostly a matter of deciding at the start what you are measuring.
Do not measure activity
The instinct is to measure hours, meetings, or documents produced. All three are actively misleading.
An advisor who spends more hours is not producing more value; frequently the opposite, since the most valuable interventions are short. A meeting cadence measures a calendar. Documents measure a word processor.
Worse, measuring activity creates the wrong incentive. It rewards visible busyness over the quiet, decisive contribution that takes an hour and saves a year.
Measure these instead
Decisions made, with a date. The clearest evidence an advisor is earning their place is a list of decisions the company made that it had been deferring. Which platform. Whether to replace or fix. What the contract will say. What risk to accept and who owns it. Not recommendations. Decisions, taken and recorded.
Run this quarterly. If the list is empty for two quarters, either the advisor is not pushing or the company is not deciding, and both are worth naming.
Risks closed or formally accepted. A risk register is only useful if items leave it. Items should be closing, or being explicitly accepted by a named person with a review date. A register that is the same size and content as last quarter means nothing is being worked.
Money found and money avoided. Contracts renegotiated, duplicate tooling retired, renewals caught before the notice window, projects that were proposed and correctly not done. This is the easiest measure to keep and the one most often not kept, because nobody writes down the project that did not happen.
Artefacts you own that did not exist before. The inventory, the register, the plan, the incident card, the evidence set that answers a client review. These have value independent of the relationship. If it ended tomorrow, what would you still have?
The awkward question to ask twice a year
What have you told us not to do?
An advisor who has never recommended against a project, a purchase or a vendor in six months is not exercising judgement, they are agreeing. Agreement is available free.
The corollary is also worth checking: what did you tell us that we ignored, and were you right? That one is uncomfortable in both directions and it is the single most informative question in the review.
When to end it
Advisory work should have a shape. Early on it is heavier, because there is a backlog of deferred decisions and undocumented ground. It should get lighter as the estate gets understood, the registers get built and the decisions get made.
If it is not getting lighter, ask why. There are legitimate answers, such as an acquisition programme or a genuine transformation. There is also an illegitimate one, which is that the arrangement has become a subscription.
A good advisor will tell you when the useful work is done, and will say so before you do. That is the strongest signal available that the relationship was real.