01 Two speeds
Stop fast. Scale slowly. That is why the review is weekly, and why it is short.
Checking five times a day hurts more than it helps.
A RevenueCat post on running user acquisition as a one-person team is written for paid campaigns, and its author is blunt: below $500 a day, make changes twice a week at most, because the budget cannot feed more. Their published week runs 6 to 8 hours across Monday to Friday, plus 15 minutes a day for pacing (is the budget spending evenly) and anomalies. That is for paid, where production and reporting live in the same seat.
UGC compresses the decision part into 30 minutes because production and creator management happen elsewhere in your week. The review only reads and decides.
Two speeds, not one. Eric Seufert of Mobile Dev Memo, on Sub Club, on new creatives: kill “as soon as it’s obviously not a winner,” which is often the same day or the next. “The winners take time to prove out, but the losers are pretty quick to prove out.”
So the review labels Stop first, because it is the fastest call, and demands the most evidence before a Scale, because that is the call that moves money.
How to see this in Airbridge
- Open Reports → Actuals, set the date range to the last 7 days.
- Add Installs, Start Trial, Subscribe, and Revenue. Group by Campaign (creator), then Ad Creative (video).
- Save it to My Reports, so next week is one click.
You'll see:
one row per video with the whole funnel, for the week that just ended, every week, without rebuilding it.
02 The 30-minute agenda
Same day, same time, every week. The order is the point.
Stop → Scale → Iterate → Watch, then the next batch, then one sentence.
| Minutes | Step | Output |
|---|---|---|
| 0 to 5 | Open the saved report. Put last week’s numbers next to this week’s, per video. | Numbers in the matrix |
| 5 to 15 | Label every video: Stop first, then Scale, then Iterate. What is left is Watch, with a week counter. (Lesson 06) | One label per row |
| 15 to 22 | Boost decisions. Scale rows still organic: the Lesson 07 plan. Boosted rows: promote or pause. | Spend changes |
| 22 to 27 | Fill the next batch, 3 to 5 tests. Iterations of winners first, empty cells second. (Lesson 08) | Production order |
| 27 to 30 | Write the sentence. Date it. Same file as the matrix. | Written record |
The matrix you carry between weeks
One table, angle by hook, with the creator and one primary metric in each cell. Rows are your Ad Group values, columns are your Ad Creative values, which is exactly why this playbook’s naming puts the angle in Ad Group and the variation in Ad Creative.
| Angle (Ad Group) | Hook A | Hook B | Hook C |
|---|---|---|---|
angle-gym-confusion | creator-maya · Subscribe 0 · Stop | creator-maya · Subscribe 4 · Scale | creator-maya · clicks 40 · Watch (week 1) |
angle-no-time-to-plan | creator-jake · clicks 212, installs 9 · Iterate (store page) | empty: next batch | empty: next batch |
angle-beginner-intimidation | empty: next batch | empty: next batch | empty: next batch |
An empty cell is a production order.
If you are boosting, keep a 15-minute daily pacing check and one automated rule in the ad platform. The RevenueCat author’s example: “pause this ad group if CPA exceeds $X over the last 3 days” (CPA is cost per action). If you are organic only, skip the daily check. The week is the unit.
03 Ask your data
Stop doing the reading by hand. Ask the report the five questions.
The prompt does minutes 0 to 15. You still make the calls.
Once you have a few weeks of history, the reading part of the review can be a question instead of a scroll. Airbridge MCP (currently in beta) is the connector that lets an AI assistant such as Claude or ChatGPT read your Airbridge reports and answer questions about them in plain language. You need a paid plan on the AI side, and your naming and events (Lessons 04 and 05) have to be in place, or the answer will be as messy as the data.
The point is not to let AI invent your next creative. It is to use your own performance history to find the pattern faster.
Ask your UGC data
Use with Airbridge MCP (beta). Works best once every video has its own link and Subscribe is flowing.
Your weekly review prompt
Other questions worth asking
- “Which creators had strong install volume but weak paid conversion?”
- “Which angles improved paid subscribers compared with the previous batch?”
- “Are any of my winning videos starting to deteriorate?”
- “Based on the last four weeks, what should I test next?”
How to see this in Airbridge
- In Airbridge, open the MCP setup page (Help Center: “Airbridge MCP”) and follow the steps for your AI client.
- Paste the prompt above into that client.
- Compare its answer with the saved report once. After that, trust it for the reading and keep the deciding for yourself.
You'll see:
the five answers in one message, drawn from the same Actuals report you would have scrolled.
04 Log the sentence
One dated line per week. That is the whole record.
“It did well” is not a strategy. A sentence with four names is.
Write it in the same file as the matrix, above last week’s line. After eight weeks you have something no dashboard gives you: the reasons. “Problem-first hook + beginner creator + visible product demo” is a pattern you can brief. “maya_final_FINAL_03.mp4 performed best” is not.
Then start over at Lesson 01 with the next batch. The review only works on volume.