How to run a weekly growth review

Guide · Frameworks · 4 min read · last verified 2026-07-27

A weekly growth review is a short, recurring meeting in which a team checks what changed in its market position, verifies what actually shipped, decides what the evidence says to do next, and makes exactly one kill-or-keep call. It is the operating unit inside a quarterly cadence: the quarter aims, the week steers. Nearly everything that goes wrong with growth meetings — drift, status theater, the slow slide into a metrics scroll — comes from forgetting that a review exists to change a decision, not to narrate effort.

The week steers, the quarter aims

The quarterly frame — the 90-day growth loop — owns direction: which positions to attack, which benchmark to lock, what done looks like at the end. That piece is about the quarter; this one is strictly about the week inside it, and the two should not blur. A week is the shortest interval at which external position changes are signal rather than noise, and close to the longest a team can run without correction before wasted work accumulates. Reviewing daily means reacting to noise. Reviewing monthly means discovering in week four that week one's plan was wrong and paying for the discovery three times over.

The thirty-minute agenda

Four segments, one hard stop.

Two rules keep the table honest. The pre-read carries the data so the meeting can carry the argument — nobody reads a dashboard aloud to a room that can read. And every segment ends in a sentence someone writes down, because a review that leaves no written residue leaves no memory either. The pre-read itself stays light: a single page of position evidence plus the list of open experiments, sent the day before, so the half hour never opens with silent reading and never depends on anyone's recollection of last week.

What moved means position, not activity

Status theater is reviewing activity instead of position, and it is the default failure state of every recurring meeting. Activity is internal and always defensible: posts published, emails sent, features released. Position is external and indifferent to effort: what buyers are asking, how answer engines describe the category, who gets cited for the questions that matter commercially. A team can win every activity week for an entire quarter while its position quietly erodes, and the review is the instrument that catches the divergence — but only if the first segment reads external evidence rather than an activity log. This is where the pre-read earns its keep: a re-measured benchmark, the kind Magrios produces on a schedule, turns the what-moved segment into a reading rather than a debate. For one worked example of a position check — tracing whether shipped content altered what assistants say — see this walkthrough of before-and-after answer evidence.

The one-decision rule

The final five minutes exist because reviews without a forcing function converge on comfortable postponement. Each week, exactly one running experiment receives its verdict: killed, kept, scaled, or rerun with a named change. One is deliberate. Zero decisions turns the meeting into a briefing; attempting five makes every verdict shallow. The verdicts accumulate in the experiment archive — the write-up discipline in how to document growth experiments is what makes each one durable — and after a quarter that archive is the most honest record the team owns of what it learned and what it paid to learn.

A useful health check: if the kill-or-keep segment keeps concluding that everything should run another week, either the experiments were launched without pre-agreed decision conditions, or the review is flinching. Both are fixable once named. Unnamed, both dissolve the cadence from the inside.

Keeping the room small and the questions standing

The review needs whoever executes and whoever decides — usually two to five people, one of whom owns the agenda and the written residue. In a founder-only company it is a solo ritual on the calendar, and it works exactly the same way. What must not vary week to week is the question set: a standing list of tracked positions, in the spirit of the five questions every growth plan must answer, so that this week's answers are comparable with last week's. Rotating questions produce unfalsifiable progress; standing questions produce a trend. Almost everything else about the meeting is negotiable — it can move days, shrink, happen over coffee. The two fatal variations are skipping it in a busy week, because busy weeks are precisely when position drifts unwatched, and letting it swell past its half hour, because a review that costs an hour will eventually be cancelled — and a cadence, once broken, is the hardest thing on this page to restore.