01 Use two signals
A UGC creative has two jobs, and they are not the same job.
Maya won TikTok. Jake won the business.
First, it has to get someone to respond. Then it has to bring the kind of user your business actually wants. Those are related. They are not the same thing.
Did the creative get people moving? Watch rate, click-through rate (CTR), installs, cost per install. The ad platform owns most of this.
What happened to those users afterward? Trial starts, paid subscribers, revenue. Airbridge owns this, per creator and per video.
| Views | CTR | Installs | Trials | Paid | Revenue | |
|---|---|---|---|---|---|---|
| Maya | 142K | 1.9% | 310 | 71 | 29 | $1,450 |
| Jake | 81K | 1.3% | 180 | 63 | 37 | $2,080 |
Maya wins every column the ad platform can show you. Jake wins the two columns that pay rent. If you only had the first three columns, you would have scaled Maya.
You need both before you move money. A creative can have a great ad signal and a great business signal, a great ad signal and a weak one, or a weak ad signal and a surprisingly strong business signal. Those lead to different actions. That is what the next section decides.
02 Four buckets
Put every live creative in one of four buckets before you decide what gets more budget.
Scale needs both signals. Stop needs neither.
| Business signal strong | Business signal weak | |
|---|---|---|
| Ad signal strong | SCALE · People respond, and the users convert at a cost you can live with. Put more behind it and keep watching the same business outcome as spend rises. | ITERATE · Cheap installs, few subscribers. Do not scale it. Find the first drop after install and test one likely cause. |
| Ad signal weak | ITERATE, keep the idea · Fewer people respond, but the ones who do are unusually valuable. Keep the angle, test a stronger hook or delivery. | STOP · Not attracting enough users, and the ones it attracts do not convert. Stop funding this version. Record the learning and move the slot. |
Not enough data → WATCH. Lesson 01’s rule has three outcomes. Watch is what you do until the rule can fire. It does not mean “leave it running forever.” It means “we do not have enough evidence yet,” and it gets a week counter.
Stop is not a failure. It is the bucket that frees a creator slot and a test slot. The learning goes into Lesson 08.
03 Find what broke
“Iterate” only matters if you know what you are iterating.
Read the row left to right. The first drop is what to change.
Treat these as diagnostic directions, not proof of cause. Views stay in the ad platform. Airbridge shows you from the click onward. Neither shows you why.
Intent exists, people vanish before the store. Check next: the path after the click. Landing page, store page, or a promise the store page does not keep.
Users arrive but show little intent. Check next: whether the creative’s promise is bringing the right users, and where they drop before the trial screen.
The creative did its job. The problem may be later. Check next: paywall, offer, onboarding. Flag it for the product side and do not blame the creative alone.
Do not stop this one because another creative has prettier ad metrics. Keep the angle, improve the opening. This is the most interesting Iterate there is.
What about cost?
Two kinds of cost, two ways in.
- Boosted or paid videos: connect the ad account you spend from and cost flows in on its own. Cost per install and return on ad spend appear for those rows.
- Organic creator fees: Airbridge cannot see what you paid a creator. If you want cost per install per creator inside the report, upload a CSV of what you paid, one row per creator per day. Open Integrations > Cost Data and click CSV File Upload. Put
ugc-creatorsin the channel column and the creator’s name in the campaign column, so each row lands on the right creator. Or keep it simple and do the division in your sheet: what you paid ÷ paid subscribers from Airbridge.
Either way, “customer acquisition cost” is a concept you own. The report gives you the denominator.
How to see this in Airbridge
- Open Reports → Actuals.
- Add Installs, Start Trial, Subscribe, and Revenue, in that order.
- Group by Ad Creative and read each row left to right.
You'll see:
the stage where each video stops producing. The columns to the left of the drop are what the creative got right.
04 Make the call
One bucket per creative, with a next action.
Scale the result, not the screenshot.
What should I do with this creative?
Answer two questions. The bucket and the next move come out.
01How is the ad signal?
Think watch rate, click-through, installs, or cost per install, whichever matches the test.
02How is the business signal?
Think trial starts, paid subscribers, revenue.
Your call
Answer both questions to get the call.