Weekly growth reviews that actually change spend

Most growth meetings review the past and change nothing. Everyone looks at the same dashboard, agrees the week was fine or was not, and the budget stays where it was because moving it feels risky and staying feels neutral.

Staying is not neutral. Every week spend sits where it was is a week of not learning. So we run growth reviews with a rule: the meeting ends with a change to spend, or a written reason why not. This is the format we use for Grow partners and for our own ventures. It takes forty minutes.

What the review is for

The review exists to make one class of decision: where next week's money goes. Not to report, not to align, not to brainstorm. Reporting happens on the page before the meeting. Brainstorming happens in a separate slot with a different set of people. Keeping those out of the room is what gets the meeting to forty minutes.

The page

• Spend, revenue, and blended cost per new customer, this week and the trailing four.
• New-customer share of revenue. If it falls while revenue holds, retargeting is eating the number.
• Per-channel cost per new customer with a confidence band, not a point estimate.
• Creative fatigue: top ads by spend with their trend over the last three weeks.
• Tests in flight: what is being tested, when it reads out, what decision it informs.
• Event health: warehouse versus platform conversion counts, with the gap.
• One line of narrative per section, written by the person who owns it.

One page. If it needs a second page, something on the first one is not earning its place. The page is the same every week so the eye learns where to look, and a number that moves is visible without a chart.

The agenda

1. Minutes 0 to 5: event health. If the numbers are wrong, the meeting is about fixing the numbers.
2. Minutes 5 to 15: the trailing four weeks. What moved, and whether we believe it.
3. Minutes 15 to 25: tests. Read-outs first, then what to launch.
4. Minutes 25 to 35: the spend decision. Proposals, objections, a decision.
5. Minutes 35 to 40: log the decision and the reasoning. Assign the changes.

The decision rule

Every review ends in one of three ways. Spend moves between channels or campaigns, spend goes up or down overall, or the group writes down why neither happened. The third option is allowed. It is not the default. A written reason has to name what evidence would change the answer next week, or it is not a reason.

In practice this produces a lot of small moves and a few big ones. Small moves are the point. A program that shifts a tenth of its budget every week toward what is working compounds. A program that waits for certainty does not, because certainty arrives after the creative has fatigued and the season has turned.

The prep

An agent builds the page before anyone is awake. The channel owners read it, edit the narrative lines, and sign off. Nobody presents in the meeting because everyone has read the page. If the page is not signed off by the start, the meeting is cancelled and the reason is logged, which has happened exactly as often as it needed to.

routine: growth-review-prepschedule: "Mon 06:00 America/Denver"steps:  - pull: [warehouse.orders, warehouse.spend, ads.platform_reports]  - reconcile:      compare: [platform_conversions, warehouse_conversions]      alert_if_gap_over: 0.05  - compute:      - blended_cac: { windows: [1w, 4w] }      - new_customer_share      - per_channel_cac: { with: confidence_band }      - creative_fatigue: { top_n: 10, trend_weeks: 3 }  - draft_narrative: one line per section, flag anomalies  - deliver: growth-review page and a summary to the channel  - require_human: sign_off before the meeting

Reading the numbers

Two habits make the numbers useful. First, confidence bands. A channel's cost per customer is a range, and a decision that only makes sense at the point estimate is a coin flip. Second, trailing windows. One week is noise for most budgets. Four weeks shows direction. We look at both and act on the four unless the one is alarming.

The number we watch most closely is new-customer share. It is the earliest signal that a program has drifted into paying for customers it already had. Blended cost per customer can look healthy for a month while that happens. New-customer share cannot.

What we do not discuss

• Creative ideas. There is a creative review for that.
• Platform changes and industry news, unless they explain a number on the page.
• Anything without a number attached.
• Last week's decision, unless the evidence it was waiting on has arrived.

The log

Every decision goes in a log with the date, the change, the reasoning, and what would reverse it. Once a month we read the log back. The pattern that shows up most often is a decision that was right on the evidence and wrong in hindsight. That is fine. The pattern we fix is a decision that keeps getting deferred for the same reason, because that means the page is missing the number that would settle it.

The performance-marketing stack playbook describes what to build. This is the meeting that runs it. Together they are most of what Grow is, and both are things a two-person team can hold. If you run your review differently and it changes spend, we want to hear how. DM @chancerobertson on X.