A marketing team sends its production partner in China a screenshot: the curve reaches 100 just as a product demonstration appears. The brief becomes “move this moment forward and make ten more.” The production request is specific, but the evidence behind it may have changed units on the way into the brief.
For brands coordinating overseas campaigns with a Chinese supply chain, this matters before a shoot is booked. A relative peak can tell the team where to look. It does not, by itself, tell the buyer which ad produces more sales or how much additional production will pay back.
What the official scale actually says
TikTok’s Top Ads Dashboard guide, updated May 2025, describes its second-by-second graph as relative performance. The official Top Ads one-pager specifies a 0–100 index: 100 marks the peak within the video for the selected metric, and 0 the lowest point. The PDF carries no publication date; both were checked on November 1, 2026.
That definition does not make 100 a 100% conversion rate. Nor does it establish that two videos reaching 100 have equal absolute performance. The cited explanation supplies no conversion from that index to orders, so an analyst should not invent one.
There is also a sampling limit. About Top Ads, updated May 2025, describes a collection of high-performing creatives authorized by advertisers. ONCE’s interpretation is that this is a useful reference collection, rather than evidence of a random sample of every advertiser’s output. These documents describe one tool; they do not establish an industry-wide shift to automated creative buying.
ONCE’s scale, comparison and action checks
This is our proposed decision framework, not a platform scoring system or a claim of measured client results. Before a reference enters a production brief, its owner should answer three questions.
Scale: what is the number measuring?
Keep the selected metric, axis definition, date range and capture date alongside the image. Label a within-video index as a relative score. Do not silently replace that unit with a percentage in a presentation. If a denominator or calculation method is unavailable, record it as unknown.
Comparison: which objects can this chart compare?
Different moments in one video can help locate a passage worth inspecting. Peaks on separate, individually scaled charts do not provide a ranking of absolute outcomes. Keep public references separate from your account’s campaign data, and establish consistent definitions and conditions before combining results.
Action: what commitment does the evidence support?
Distinguish reviewing a passage, commissioning a test variant and expanding production. A relative peak can justify closer inspection and help formulate a test. Expanding production requires relevant business results and cost information from your own operation, assessed with an appropriate comparison design.
| Evidence available | Reasonable next step | Unsupported leap |
|---|---|---|
| A high point in one video | Inspect the surrounding sequence | Attribute incremental orders to that shot |
| Three videos each reaching 100 | Develop separate creative hypotheses | Rank their absolute conversion performance |
| Consistently defined account outcomes and costs | Evaluate whether to retain, revise or reshoot | Declare a universal rule across products and audiences |
Brief a comparison the production team can actually make
Consider a hypothetical home-accessory brand sourcing from China. Its team notices a reference curve peaking near a storage product’s reveal. The usable observation is that a relative high point appears near the reveal. “The reveal caused sales” is a separate causal claim that this chart alone does not establish.
The brand could commission two original openings using its own product: one establishes the household problem before showing the result; the other shows the same genuine result first. Keep the offer, later proof and call to action as consistent as practical. Specify the intended comparison and review date before release.
Do not also change the price, presenter and landing page, then credit the difference entirely to the reveal. Review what happens before the peak too: a reaction, an offer or an earlier setup might be relevant. If the resulting account data cannot answer the business question, the hypothesis remains unresolved. It should not become a permanent production rule.
Buy a usable research handoff
A supplier’s reference card should contain the source and capture date, metric and scale, observed passage, unknown conditions, proposed original test and reusable shots. Acceptance should depend on whether that card changes a script or a test decision. A large folder of screenshots is a weak substitute.
For an ongoing creative partnership, keep the eventual result beside the original hypothesis, including a negative or inconclusive result. Public references generate possibilities; authorized account and sales records answer the brand’s own performance question. Maintaining that distinction helps production work build a reusable method over time, without turning other advertisers’ examples into invented client evidence.
Teams preparing a production brief can explore ONCE’s video marketing articles. Start the next reference review with a small discipline: write the unit next to 100 before approving additional versions.
Research scope
This evergreen analysis uses the three official sources linked above, verified November 1, 2026. It is not a report of a same-day product release. No client advertising account was accessed, no actual conversion rate was calculated, and no campaign experiment was conducted for this article. Examples and proposed actions are illustrative.
