Our packages

The review that decides: how to run the meeting so that it is worth holding

A review that decides nothing is a report read aloud. The hour is the smallest of three parts, and the only one anybody prepares.

Published on 5 August 2026 — Algeria Agency

The article beside this one says that at this level the report becomes a decision meeting. That is the real difference between a service that is managed and one that is merely delivered, and it is also the easiest thing to let decay without anybody noticing.

A review always decays the same way: it becomes a presentation. The supplier shows what was done, the numbers go up or down, everybody nods, the meeting ends on time, and nothing has been decided. It looks successful and it has cost five people an hour.

What separates a useful review from a presentation is not in the hour itself. It is in what is sent beforehand, in who is present, and in what is written afterwards — three things of which only one is usually organised.

This article gives the shape of the three parts, the agenda that produces decisions, the three numbers that are enough, the decision nobody wants to take, and why we will never publish what a review "returns".

A review that decides nothing is a report read aloud

The test is brutal and takes a minute at the end of the meeting: name the decisions taken. Not the information shared, not the progress noted — the things that will be different next week because that hour happened.

In most reviews the honest answer is none. That is not a sign of bad faith: it is what a meeting built as a presentation produces, where the supplier talks for eighty per cent of the time and the client asks questions of comprehension.

A decision has three recognisable marks. It changes what somebody will do. It has a named owner. And it could have gone the other way — if no other option existed, it was not a decision but a piece of information.

That last criterion sorts most efficiently. "We continue the campaign" is a decision only if stopping was on the table. Said with no alternative, it is a description of what was going to happen anyway.

The consequence is not that more should be decided. It is that a review has to be built so that a decision is possible inside it, and a presentation is not — because by the time the options appear, there is no time left.

The three parts, and the shortest is the one everybody prepares

A review is made of three things of very different sizes: what is sent beforehand, the hour itself, and what is written afterwards. The second is the shortest and it is the only one that gets organised.

The work beforehand is the longest and the most profitable. That is where the numbers are gathered, the options are framed, the questions are written. Done properly it transforms the hour: people arrive knowing what will be discussed and what is expected of them.

The hour itself should be the shortest part of the arrangement, and it should be used almost entirely for settling things. Anything that can be read is read before; anything that can be written is written after. What is left in the middle is what needs several people in one room at one time, and that is not much.

The work afterwards is the shortest of the three and the most often absent: a list of decisions, each with an owner and a date. Without it the next review opens with "where were we", which consumes a quarter of the hour and lets the same points slide from month to month.

The overall rule is therefore counter-intuitive: to improve a review, you almost never improve the meeting. You lengthen what comes before it and add what comes after, and the hour shortens by itself.

What goes out forty-eight hours ahead

A pre-read is not a report. It is a short document whose function is to make reading unnecessary during the meeting, and it fits on a page and a half.

It contains four things. The period’s numbers, with no commentary. What was done, as a list. The decisions expected, framed as questions with their options. And what is blocked, with the name of whoever unblocks it — often somebody on the client’s side.

The third is what changes everything. A question put with two written options lets somebody think about it while walking, mention it to a colleague, arrive with a position. The same question asked for the first time in the meeting produces either a deferral or a decision taken in four minutes by people encountering it.

Forty-eight hours is a chosen number rather than a ritual: long enough to be read, short enough for the numbers to still be true. Sent the night before, the document is not read; sent a week ahead, it is read and then forgotten, and has to be reread in the room.

Finally, the pre-read has a side effect nobody anticipates: it disciplines whoever writes it. Framing a question with its options forces you to have a position, and a supplier who has to write their recommendations forty-eight hours ahead produces better ones than the supplier improvising in front of a shared screen.

Who should be in the room, and who should not

A review needs three roles and not one more: somebody who can decide, somebody who knows the ground, somebody who did the work. Sometimes that is three people and sometimes two; it is never eight.

The decider is the only indispensable role. A review without one is not a review: it is a preparation meeting, which is sometimes useful, and it should be called by that name so nobody expects decisions it cannot take.

The person from the ground is the one most often missing, and their absence costs in a particular way: they alone know what customers actually say on the telephone. Without them, decisions get taken on numbers, which are always behind what the counter already knows.

You also need to know who should not be there. People present "for information" lengthen the meeting and make it more cautious: nobody settles a difficult question as readily in front of an audience. They are far better served by the decision record sent afterwards.

A review of three that decides beats a review of eight that informs, and moving from the second to the first costs no budget — only sending the record to the people no longer invited.

The agenda that produces decisions

An agenda that opens with the numbers ends with the decisions, and the hour runs out before it gets there. It is mechanical: numbers are comfortable and decisions are not, so time flows towards the first.

The order that works is the reverse. Open with the decisions expected, which were in the pre-read. Settle them. Then look at the numbers, asking only what they change for the next ones. Then close on what is blocked.

Each decision point is handled in three short moves: the question, the options with their consequences, and the position recommended by whoever did the work. A supplier presenting three options without recommending one is handing you back your own work.

You also need an exit rule. If a decision cannot be taken for want of information, it is not deferred "to the next meeting": somebody is named to bring what, by when, and that line goes into the record. A deferral with no name and no date returns identical the following month.

Finally, a review must be allowed to be short. A month with nothing to settle is a month when the meeting lasts twenty minutes, and it should be ended — an hour filled because it was booked teaches everybody that their presence is decorative.

The numbers: three are enough, and not the dashboard’s

A dashboard holds thirty indicators because the tools supply them free. A review uses three, and choosing the three is a decision taken once for the year rather than in the room.

The criterion is simple and it eliminates almost everything: a number belongs if there is a decision you would take differently depending on its value. The rest is general knowledge about your own business, and it can be read in the pre-read by anybody who wants it.

The three that survive that test are almost always of the same kind: how many enquiries arrived, what they cost, and what became of them. The third is the one most reviews do not have, because it lives at the client rather than at the supplier.

That is a point worth stating plainly. A supplier knows how many enquiries it produced; it does not know how many became customers unless somebody at your end tells it. A review missing that judges the work on its less important half.

You also need a rule about variation. A number moving ten per cent in a month means nothing for most Algerian businesses of this size, because the volumes are too small to separate a change from a fluctuation. Deciding not to comment on a single month’s variation saves a quarter of an hour and avoids decisions taken on noise.

"Everything is fine" is the most dangerous meeting

A review where everything works is the one that decays fastest, and it does so invisibly because nothing is wrong. It is the month cancelled for a busy calendar, then the next one, and six months later the relationship has become a monthly invoice.

What those months should produce is not a record of satisfaction. It is a precise question: what would make us stop the thing that is working. A profitable campaign has a limit of scale, a channel has a season, a page that converts eventually tires its audience.

It is also the right moment for the one question difficult reviews never allow: what should we be trying that we are not. When nothing is on fire there is room for a bounded experiment; when something is on fire there is none.

It should finally be used to check what is not being looked at. A service that works concentrates attention on what works, and the things that decay slowly — a response time, a listing going stale, a product feed no longer current — never announce themselves.

The practical rule: a month with no problem shortens the review and changes its agenda, it does not cancel it. Cancelling is what turns a managed service into a subscription, and nobody decides to make that change — it happens to whoever did not refuse it.

The decision nobody wants to take: stopping something

The most useful decision a review can produce is almost always a stop: cutting a channel, dropping a format, ceasing to produce something nobody reads. And it is the decision meetings take least.

The reasons are human and worth naming. Stopping means admitting the last six months of budget did not work. It also removes work from somebody, internally or at the supplier. And it has no visible benefit: nobody congratulates a business on a campaign it did not run.

The way to make that decision possible is to prepare it rather than discover it. One line of the pre-read, every quarter, asks the same question: what would we stop if we had to remove one thing. Asked regularly, it stops being an accusation.

The stop also has to be reversible and dated. "We cut it for two months and look again" is a far easier decision to take than "we stop", and it produces the same result nine times out of ten — because most things that get suspended are missed by nobody.

Finally, a stop should appear in the record exactly as a launch does. A business that cannot name a single thing stopped at the end of a year has not been deciding for twelve months: it has been adding, which is the only direction a service drifts in on its own.

What the supplier must bring that you cannot produce

At this level you are paying for something specific, and you should be able to check that you are getting it. It is neither the report nor the meeting hour: it is three things a business cannot produce about itself.

The first is comparison. A supplier working for several businesses knows what is normal and what is not, and that information exists nowhere at your end. It has to be given without naming anybody, and a supplier who never gives it is keeping back half of what justifies its price.

The second is a recommendation. Not options: a position, argued, that you can refuse. A supplier who systematically presents three equivalent routes is protecting itself from being wrong, and handing you back a decision you paid it to work through.

The third is bad news brought early. A channel running out of steam, a test that showed nothing, an assumption that does not hold: that is said in the month it is seen, not in the quarter when it becomes undeniable. It is the hardest thing to demand and the simplest to check — look at whether your supplier has ever told you about a failure before you noticed it.

Those three are the criteria for a review at this level. If all three are missing for two quarters, what you are buying is a report, and a report can be produced without a meeting.

The decision record, and the channel between reviews

The record is the cheapest part of the arrangement and the most often absent. It is ten lines, sent the same day, and it contains decisions only: what, who, when.

It does not replace minutes and should not resemble them. Minutes narrate the meeting; a record registers what changed. The first is never reread, the second is the first item of the next review.

It also protects in the only direction that matters: against mutual forgetting. Two months on, nobody remembers who was to supply the photographs, and the discussion that follows costs more than the task. One line with a name settles that permanently.

Between reviews there has to be one channel and only one. The usual failure is that decisions get taken outside it — in a message late at night, in a corridor, in a conversation between two people — and never enter the record. The result is a review discussing a state of the world that is no longer the right one.

The rule that works is short: what is urgent goes through the direct channel and is carried into the record at the next review. What is not urgent waits for the review. A service where everything is urgent has no review, it has a queue of messages.

Why we do not publish what a review "returns"

The question is legitimate, particularly at this price: what does this monthly meeting return. We give no figure, and the reason lies in the nature of what a good review produces.

What it most often produces, when it works, is a stop. A channel cut, a project not launched, a spend that did not happen, an idea dropped before it was built. It is the most profitable decision and it is the previous section.

But nothing anywhere counts the campaigns that were not launched. There is no register of avoided spending and there cannot be one: a thing that did not happen leaves no invoice, no accounting line, no trace in any tool. The most important effect of this arrangement is invisible to every instrument, ours included.

The consequence is that a published figure on this subject could only measure the additive decisions — launching, producing, increasing — which is precisely the half that drifts on its own. It would value the review for what it does least usefully, and a supplier optimising that figure gains one more reason never to stop anything.

What we look at instead is an annual question you can answer alone: name the things you stopped this year. If the list is empty after twelve reviews, the meeting has not been deciding — it has been informing, and informing is done in writing, without occupying three people for an hour a month.

What we do, and what we refuse to do here

We send the document forty-eight hours ahead, with the numbers uncommented, the decisions expected framed as questions and options, and what is blocked with the name of whoever unblocks it — including when that name is at your end.

We open the meeting with the decisions rather than the numbers, we recommend a position on each, and we send the record the same day. Once a quarter the pre-read carries the line "what would we stop", and it is there whether the period was good or bad.

We refuse to hold a review without somebody who can decide. We will gladly hold it as a preparation meeting and call it that, because a review without a decider produces deferrals and gives the impression that the arrangement is working.

We refuse to present three equivalent options without recommending one. It is comfortable for a supplier and it hands you back the work you paid for; if we have no position we will say so, and say what would have to be known in order to have one.

And we refuse to put a figure on what the review returns to you. The previous section says why: its most useful output is a thing that did not happen, and a figure that can only count additions would give us one more reason never to stop anything. We prefer the end-of-year question, which you ask and we answer together.

Frequently asked questions

How often should this review be held?

Monthly in most cases, with a longer quarterly one that asks the stopping question. The rhythm matters less than the rule: a month with nothing to settle shortens the meeting, it does not cancel it. Cancelling is what turns a managed service into a subscription.

Who should attend?

Three roles: somebody who can decide, somebody who knows the ground, somebody who did the work. The decider is indispensable — without one it is a preparation meeting, which is sometimes useful but should be called that. People attending "for information" are better served by the record.

How many indicators should we track?

Three, chosen once for the year. The criterion eliminates almost everything: a number belongs if there is a decision you would take differently depending on its value. And do not comment on a single month’s variation — at these volumes it does not separate a change from a fluctuation.

What do we do when everything is fine?

Shorten the meeting and change its agenda rather than cancel it. Those months serve three purposes: asking what would make us stop what is working, opening a bounded experiment while there is room, and looking at what decays slowly — a response time, a listing going stale — because none of it announces itself.

What does a review actually return?

We do not publish that figure. What a good review most often produces is a stop — a channel cut, a spend avoided — and nothing anywhere counts the campaigns that were not launched. A figure could only measure the additions, which gives one more reason never to stop anything.

How do we tell whether ours is working?

Ask at the end of the year: what have we stopped. An empty list after twelve reviews means the meeting informed rather than decided — and informing is done in writing, without occupying three people for an hour a month.

Where we come in

The note sent the same day is what separates a decision from an intention, and it runs to ten lines.

We hand you the figures two days ahead, with no commentary around them.
We put every expected decision as a closed question.
We open each quarter with what should stop, before what should grow.

If your review exists to inform rather than to decide, do not give it to us: we would make nothing better of it than an email.

See the Premier page Start with the free audit

Let us talk about your project

A free audit, no commitment: we look at your online presence and tell you what is holding it back.

Request the free audit Write to us