AI Creative Analysis
We ran an AI analysis of TikTok beauty content performance across thousands of videos. The clearest finding was not a list of best practices. It was proof that best practices break the second you move from one product category to the next.
There is a comforting idea floating around marketing teams that beauty content has a formula. Hook in the first three seconds. Show the product. Keep it authentic. Talk about benefits. Brief enough creators on that and the numbers should follow.
The numbers do not follow. We pulled a large volume of TikTok beauty content into Aggero Insights and let the system read every video the way a person never could, frame by frame, audio cue by audio cue. The dataset covered three makeup categories that most brands treat as one bucket: complexion, lips and eyes. Same platform. Same kind of audience. Same creator playbooks.
What came back was not a clean checklist. It was a map of contradictions. A creative choice that lifted complexion videos by double digits dragged lip videos down by almost as much. The thing that worked depended entirely on what part of the face the creator was talking about. That is the part nobody briefs for, and it is the reason so much beauty content underperforms despite following every rule in the deck.
Nielsen and NCSolutions have spent years putting hard math behind why this matters. Creative quality drives roughly half of all incremental sales from advertising, more than targeting, reach and recency combined, and strong creative can account for up to 89 percent of digital sales lift. The creative is the lever. Most teams just cannot see which part of it is moving.
Across every category we looked at, the fastest way to lose a viewer was to open with a product demonstration. This is one of the clearest patterns in TikTok beauty content performance. The instinct makes sense. You have a product, you want to show it working. The audience reads it as an ad and scrolls.
But the size of that penalty was not constant. It got dramatically worse depending on the category, which tells you the problem is not "demos are bad," it is "demos are bad in proportion to how much the audience came for a story instead of a spec sheet."
Same underlying mistake, three different price tags. A brand that copies a "show the product" template from a generic playbook pays the highest version of that penalty in the category where the audience is least tolerant of it, and never finds out why the video died.
This is the finding that should change how you write briefs, and it is where most analyses of TikTok beauty content performance stop short. We looked at vibrant, high saturation color palettes, the kind of bold visual energy most beauty teams default to because it feels native to the platform.
In complexion content, vibrant palettes were one of the strongest hook drivers we measured. In lip content, the exact same cue was a drainer. Not neutral. Negative. The lip audience rewarded calm, realistic, neutral palettes and actively disengaged from the loud ones.
A 66 point gap on a single creative variable, driven by nothing except which product was on screen. If your brand runs both complexion and lip lines off one creative template, half your catalogue is being briefed against itself.
The pattern repeated on emotional delivery. Complexion content rewarded happy, upbeat openings. Lip content did the opposite, rewarding neutral, calm authenticity over performed enthusiasm. Eye content leaned toward slow visual pacing and contrast heavy framing, where a striking light and dark composition was the single strongest hook driver in that entire category at plus 86 percent.
None of these are wrong. All of them are wrong in the wrong category. The creator who brings high energy to a lip video is doing exactly what worked on the complexion shoot last week, and watching the result tank for reasons the dashboard will never explain.
For all the contradictions, a few signals were stable across all three categories, and they are worth treating as close to universal for beauty short form.
So the universal rule is small: lead with a human and a reason to keep watching, not a product. Everything past that point is category specific, and treating it as universal is exactly where the money leaks out.
A human reviewer can watch fifty videos and form a hunch. They cannot watch thousands and hold every visual, audio and structural variable in their head at once, then separate the complexion results from the lip results from the eye results. The flip we found on color palettes does not show up in a manual review. It does not show up in a views dashboard either, because the dashboard tells you a video underperformed, never which frame caused it.
That is the gap. Brands are producing far more content than anyone has the capacity to actually read, so they approve it blind and measure it after the fact. Real TikTok beauty content performance lives inside the videos, in combinations no spreadsheet of engagement rates will surface.
Stop writing one beauty brief. Write a complexion brief, a lip brief and an eye brief, and let each one contradict the others where the data says it should. Brief vibrant energy into your foundation content and strip it out of your lip content. Push neutral calm into lips and save the contrast heavy drama for eyes. Keep a face in every thumbnail and a story in every opening, regardless of category.
The brands that win the next year of beauty content are not the ones with better creators or bigger budgets. They are the ones who treat TikTok beauty content performance as a category by category question, stop guessing which creative element is moving the number, and start measuring it before the spend goes out the door.
By breaking each video down into its underlying creative elements, the hook, the visual composition, the audio messaging, the pacing, and measuring how each one affects engagement. Instead of looking at views and engagement rate after the fact, this kind of analysis identifies the specific drivers and drainers inside the content itself, which is what actually decides whether a video performs.
Because they treat beauty as one audience. Our analysis showed the same creative cue can be a strong driver in complexion content and a drainer in lip content. A rule that holds in one category can actively hurt you in another, so a single set of best practices applied across a full product range will underperform.
Not entirely, but they should almost never open the video. Across every category, demo led openings were among the largest engagement drainers because viewers read them as ads. Leading with a person and a reason to keep watching, then earning the demo later, performs far better.
Enough that the category splits stay statistically meaningful. This analysis ran across thousands of videos and a large pool of creators. The contradictions between categories only become reliable at that scale, which is exactly why manual review misses them.
Related reading: the Aggero blog, real case studies, and the video content analysis 101 guide.
Internal figures are drawn from Aggero's AI analysis of an anonymized short form beauty content dataset spanning complexion, lip and eye categories. Percentage values represent performance relative to the dataset average. External creative effectiveness figures: NCSolutions, Five Keys to Advertising Effectiveness (2023); Nielsen.
