The quarterly review: holding a provider to what they signed

The companion article is about buying: what is really bought when time is bought, when the clock starts, what a contract must contain, and how to change provider without being held hostage.

Between those two moments — signing and leaving — there are two or three years nobody writes anything about. That is where it is all decided, and it is this page’s subject.

What makes it necessary is one observation: a maintenance contract does not fail all at once. It thins. Response times lengthen by a few hours each quarter, the same faults come back, and nobody can say when it started.

The answer is four numbers taken from **your** side and a thirty-minute meeting each quarter. And one point that decides everything: if you measure nothing yourself, the only available version of events is that of the person being assessed.

Of those four numbers, reopenings are the hardest to dress up and the richest in information, and they deserve their own section.

An incident closed and then reopened within the week says something precise: what was done addressed the symptom rather than the cause. That is a technical fact rather than an accusation, and it reads without any particular knowledge.

The reason the figure is reliable is mechanical: closing is decided by the provider, reopening is decided by you. No presentation can make an incident you reopened yourself disappear.

What to look at is not the rate but the distribution. Three reopenings spread across three different subjects describe a team moving fast; three on the same workstation or the same server describe a problem nobody has yet diagnosed, and it is the one line in the review that deserves an agenda item on its own.

One variant to watch for, quieter: the incident that is not reopened but reopened **under another name**, a week later, by somebody else in the office. It is the same problem and it shows up as two resolved incidents. That is what the fourth number exists for.

Everything above rests on a notebook you keep, and it has to be small enough to actually be kept. Five columns, one line per incident, forty seconds of writing.

The date and time of the report. Who reported it. What was not working, in one sentence of your vocabulary rather than theirs. The time of the first real response. And the "seen before?" column from section 3.

What is not in it matters as much: no diagnosis, no cause, no judgement on the quality of the visit. Those three require expertise, take time, and are exactly what makes a register get abandoned after six weeks.

Who keeps it is the real question, and the answer is not "everybody". One person, the one who already takes the internal calls — often the office manager, or the person the support article calls the internal relay. It is a few minutes a week added to a role that already exists.

And it lives on paper or in a simple shared file, next to the person who keeps it. A ticketing tool bought for this is a good idea that fails at this size: it needs administering, it produces reminders nobody reads, and within a quarter it becomes a second system alongside which the real exchanges continue by phone.

A review sometimes produces a request to adjust, and it is better to know in advance what moves easily and what does not, so as not to spend goodwill in the wrong place.

What moves easily: scope. Adding the new site’s workstations, removing the three machines that were scrapped, including the backup that was outside the contract. Those are lines, they are recalculated, and both parties gain.

What moves moderately: the stated response times. A provider can commit harder on a narrower scope — the three critical workstations rather than the twenty — and that is nearly always the right trade to propose, because it asks for no extra resource, only a written priority.

What does not move: the price, at equal scope and mid-year. Asking damages the relationship for a gain that will not come, and there is a moment provided for it, which is renewal.

And one thing that never renegotiates and is asked for at the first review: that the administration credentials for your equipment sit in a document at your premises. That is not a commercial clause, it is the condition that makes every other discussion possible — and it is the neighbouring article’s section 11, seen from inside the relationship rather than at its end.

What we refuse first, when we are the provider being assessed: producing the numbers that grade us. We supply our record of visits, which is a fact, and we do not manufacture the indicator of our own quality — the client’s register carries that, and we gladly help set one up for somebody whose provider we are not.

We also refuse to hold, alone, the agenda of a review that concerns us. A meeting whose items are written by the party being assessed is a meeting where the absent subjects are never absent by accident.

And we refuse to conclude that an incumbent provider should be replaced without having looked at section 9. It is more profitable for us to conclude the opposite, which is precisely why it is written here.

What we do gladly, and it takes half a day: opening the five-column register with whoever will keep it, defining the four numbers, writing the template agenda, and sitting through the first review in silence so as to tell you afterwards what was not asked.

And one thing to do today without us: open a notebook, write the five columns at the top of the first page, and put it next to the person who already takes the internal calls. In three months you will have the one document nobody else can produce in your place.