A 21,300 follower pharmacist beat a 3.57 million subscriber channel

We ran a real brief through Virlia: a fragrance free barrier repair moisturiser for people whose skin is genuinely reactive. Find creators who talk honestly about damaged skin barriers, show bare skin rather than filtered results, and explain ingredients instead of listing products.

Sixteen videos came back from three search angles. Four creators were watched frame by frame. The shortlist put Dr. Nas, PharmD, a TikTok pharmacist with 21,300 followers, above Doctorly, a YouTube channel with 3.57 million subscribers.

Why the smaller creator won

Not because small is better. Doctorly scored 100 on brand safety and 90 on fit, which is excellent. The pharmacist scored 88 on fit against a brief that asked for ingredient level honesty, and the deciding factor was what the frames showed about how each explains actives on camera.

A follower count cannot express that. Neither can an engagement rate, a category tag, or an audience demographic breakdown. Those are the four things every creator platform sorts by.

The judgement marketers actually make is formed by watching. Every platform skipped it because reading a number is cheap and watching a video was not.

What the frames carried

Two creators in the same category can look nothing alike once you see them. One video was studio lit with two board certified dermatologists in scrubs, a branded title card, and a labelled clinical photograph cut in mid video. Another was a bare face on a sofa with a clip on microphone, ambient room light, no cuts to product, and the same face across all thirty six sampled frames.

Both are skincare channels with similar engagement rates. A metadata tool sees two rows that look the same. Only one of them matches a brief asking for unfiltered, and knowing which is the entire job.

The part that surprised us

Watching turned out to be the cheap part. Frame analysis is about seven percent of what a run costs while consuming most of the tokens, because vision models built for volume are an order of magnitude cheaper than the reasoning models that weigh the result.

Which means depth is close to free. Watching more of each video barely moves the bill. The expensive part is reasoning about more creators, which is the thing worth metering.