UGC Handbook / Lesson 11 of 14 / Chapter IV: Measure & scale

Don't boost the video with the most views

300,000 views. 174 likes. 8 comments.

That is the scorecard on one of Jay's organic videos. Jay runs Nomad Table, a solo-travel app with over a million downloads, and today he manages about 60 creators who generate 44 million views a month. On the Superwall podcast he called that ratio "horrible," and gave the mistake a name: the view trap.

If you had boosted that video because it had the most views, you would have paid to amplify what the interview host called clickbait "that gets great views, but has zero ability to convert users."

The view trap

Here is what the 300K video looked like. Jay is in a hostel, zooms in on a stranger, and the caption says "so this random girly in my hostel just showed me this and I'm never staying in a hostel again lol." Then the video cuts to his app.

The hook promised something gross in a hostel. The audience stayed to see it, the algorithm read that watch time as a good signal, and the views piled up. Then they saw an app instead, and almost nobody liked or commented.

Jay's own diagnosis: "It's a good hook, but I'm not delivering on the hook very well cuz it's just a clickbait thing." Views measure whether people stopped scrolling. They say nothing about whether people wanted what they saw.

The Coconote version of the same trap. Lesson 01 covered the extreme case: a Coconote side project got 41 million views and 4.5 million likes on a single video and still produced almost no paying users, because it was framed as a toy. Attention and intent are two different things, and views only count the first.

The number that separates them. Drew Levan, who helped build the UGC marketplace Sideshift, gave the working definition in a Superwall interview: engagement rate = (likes + comments + saves + shares) ÷ views. His threshold: anything above 7.5% to 8% is good, and above 12% "it'll start to go very very viral generally."

Run the math on Jay's video. 174 likes and 8 comments on 300,000 views is at least 0.06% on likes and comments alone (saves and shares were not shown). Jay's video ran on Instagram and Drew's thresholds come from TikTok Spark work, so the comparison is directional, but 0.06% is not close on any platform.

Three numbers decide the budget, and views is not one of them

When you decide what to put money behind, you are reading three signals from three places. "Boosting" here means TikTok Spark Ads or their equivalents: paying to promote an existing organic post from the creator's own account, rather than uploading a new ad.

1. Engagement rate, on the platform. This is the first filter, and the fastest to read. Drew's team ranks candidate videos in a fixed order: comment sentiment first, then engagement rate, then average views. Views come last on purpose.

Comment sentiment means what people write. Drew's example: when the comments say "what is that app?", the video "generally will do really well on Spark conversion wise."

2. Down-funnel counts, in your tracking. A video can clear 7.5% engagement and still send nobody to the store. That is why the second filter is clicks → installs → Start Trial per video, which is only visible if every video carries its own link (Lesson 03 sets this up).

A RevenueCat blog post on creative testing adds a warning here. If you optimize for an upper-funnel action like trial start, some creatives will show a low cost per acquisition (CPA) but "very poor conversion to paid subscription afterward," often because the algorithm over-delivers to 18 to 24 year olds who try apps and rarely subscribe. Lesson 12 follows the funnel past trials for exactly this reason.

3. Spend absorption, in the ad manager. Once a video is boosted, the platform tells you what it thinks. The same RevenueCat post: winning creatives "typically receive 80 to 95% of daily spend," and, when several creatives share one ad group, "if a creative receives less than 50% of the budget after two days, treat it as a likely loser or false positive."

Two days in, the budget split answers for you.

Boost rules, with the thresholds attached

The rules below come from Drew Levan's Sideshift team (Spark Ads on TikTok) and the RevenueCat creative-testing post (Meta and TikTok paid ads).

Wait before you boost. Drew's rule: "You should not preemptively spark a video. If organic is ripping, hold off as long as possible. Give it like a day or two."

Boosting a video that is still climbing organically just pays for views you were about to get free.

Boost what is already outperforming its own account. Drew's team sparks a video when it performs "three to five times better than the average viewership on an account." The stated side effect is that it "actually makes the account perform better on organic" afterward.

Start with a $20 test. Drew's first move on any video above 7.5% to 8% engagement is "$20 behind it," on the condition that CPMs (cost per thousand views) are "low enough to where you feel comfortable." Sideshift's own runs are bigger: "we usually start with $50 to $100 a day on a video," left running a week at a time.

Kill on engagement, not on views. His team's rule of thumb: "If the engagement rate drops below 5%, we kill the spark." Drew also observed that Spark CPMs typically stay flat until engagement decays to roughly 5%: the same line, from the cost side.

Kill on spend share. RevenueCat's rule for paid campaigns where several creatives share one ad group: under 50% of budget after two days means likely loser or false positive. Rotate it out and put the slot on the next candidate.

Do this before your next boost

1. List every video from the last 7 days with views, engagement rate, and the top 3 comments in one column. Sort by engagement rate, not views.
2. Add the Airbridge row for each. Clicks → installs → Start Trial per Ad Creative. Any video with high engagement and no clicks gets flagged: the content works, the CTA (call to action) doesn't (Lesson 04).
3. Classify each video as Scale / Iterate / Watch / Kill using the table below. Only Scale candidates get money.
4. Start with a $20 test on each Scale candidate, after it has had at least a day or two of organic run. If it holds, move to $50 to $100 a day, and put boosted videos in the same ad group so the spend split means something.
5. Check on day 2. Below 5% engagement, stop. In a shared ad group, a video taking under 50% of the budget after two days, stop that one. Otherwise leave it a week and recheck.

Then complete this sentence: