Open the TikTok Creative Center Top Ads dashboard, pick an ad, then choose See Analytics. The second-by-second chart shows how the selected metric varies through the video. To turn that curve into an editing decision, replay the second before a change, the moment itself and the second after. Write down what viewers could see and hear across all three.
If you came looking for a TikTok ad library because you need to look up an advertiser's disclosure, start with the Commercial Content Library guide. Top Ads serves a different research question: which moment in a selected creative is worth studying?
Choose a metric before you pick a video
Imagine you're cutting a product demo and the first shot takes four seconds to show what the product does. Start with a region and campaign objective relevant to your brief, then look for examples where the payoff arrives sooner. A six-second view rate sort makes sense when you're studying how a video holds attention into the demonstration. A click-through rate sort fits a question about what viewers see before deciding to visit.
These are two different decisions. The dashboard sort helps you choose an ad. The metric on the detail chart helps you choose a moment inside that ad. Write down both labels. Otherwise it's easy to select a video by six-second view rate and accidentally describe a point on another curve as “where people stayed.”
TikTok documents region, industry and objective filters, the available sorts, and a limited selection when browsing logged out. If your selection is small, sign in before generalizing from the handful of videos you can see. TikTok's Top Ads dashboard guide.
Read the picture and sound on either side of the peak
Watch once with the graph out of your line of sight. Where do you first understand the product? When do you first see evidence that it works? When does the offer appear? Then inspect the curve and replay the section you marked. Listen for a spoken change as carefully as you look for a cut. An offer can arrive in the voiceover a second before the price card appears.
Here's a fictional example from a wet-vacuum demo. The timestamps and sequence illustrate how to take notes. They are not measurements from an advertiser.
| Time | Picture | Sound | What to mark |
|---|---|---|---|
| 0:02 | A muddy stripe fills the frame | “One pass.” | The promise is audible before the tool moves. |
| 0:03 | The nozzle crosses the stripe | Motor starts | The cut and the sound change together. |
| 0:04 | A clean strip appears behind the nozzle | Motor continues | First visible proof of the promise. |
Suppose the selected curve rises around 0:04. “The clean strip is visible here” is a useful observation. “The motor sound caused clicks” would require a test that isolates sound. Even the strongest-looking point can't separate a new camera angle, a sound cue and a product reveal that land within a second of each other. If the chart is flat, record the first clear proof frame anyway. A good reference can answer a craft question without a dramatic peak.
Compare the same creative decision across two ads
Suppose your second reference is a twelve-second clip that opens on a clean carpet, then cuts back to the spill. Where is the first proof frame? For a viewer who has never seen the machine, a picture of a clean carpet alone tells them little. Mark the first moment they can connect the nozzle, the spill and the clean strip. That might be several seconds after the opening image.
Line up the two ads by that moment. In the first demo the viewer sees the problem, hears a promise, then sees the result. In the second, the viewer sees the result and has to wait for the explanation. Rewatch both with the sound off. If the second sequence becomes hard to follow, write down the missing information: “The voiceover is the only cue that this is the same patch of carpet.” An editor can act on that detail by keeping a distinctive tile edge or piece of furniture in both shots.
Keep the chart readings within their respective videos. Two similarly shaped peaks give you places to inspect. Their height alone doesn't tell you how many people clicked each ad. A useful comparison to bring back is the order of the shots and what the viewer could understand at each point.
What if the biggest change happens after the reveal?
Replay the interval from the first clean strip through the next spoken line. Perhaps the demonstrator turns the nozzle toward the lens just as the price appears. You now have two plausible explanations to investigate: the close-up makes the mechanism clearer, or the price makes the offer concrete. Pick the explanation your production team can isolate. For the price question, export two versions with the same close-up and audio and move only the price caption.
Sometimes the trace changes gradually and there is no single frame worth circling. Mark a short interval and describe what persists through it. “From 0:04 to 0:07 the camera stays wide enough to see one uninterrupted pass” gives the editor more direction than “engagement goes up here.” It also suggests a different experiment: keep the opening and compare one uninterrupted demonstration with a cutaway midway through the pass.
Turn the observation into one edit you can test
For the wet-vacuum example, brief two cuts of your own footage. In A, open on the muddy stripe for two seconds before the cleaning pass. In B, start midway through the same pass so the clean line is visible in the opening frame. Use the same voiceover, offer, end card and runtime. Fill the time freed in B by extending the final result shot. Record that timing difference so the editor can reproduce both cuts.
If the question is whether viewers stay long enough to see proof, use the same early-view measure for both variants in your campaign reporting. If attention improves but click-through falls, watch the transition into the offer before commissioning a different opening. The reference pointed to a moment worth changing. Your campaign supplies the evidence for keeping the edit.
Give the editor a timestamp they can use
A handoff that says “Make it like this TikTok” leaves the editor guessing. Send the source link with one instruction they can execute:
Reference moment: 0:03–0:04, first visible clean strip. Edit to try: open on the reveal in our own demo. Keep: the spoken claim and final offer. Question: do more viewers stay through the demonstration?
Keep the dashboard filters, sort and chart label with the link. If your team collects examples across platforms, organize the references in Denote and group those that answer the same editing question in a creative swipe file. A folder called “first frame shows the result” gives the next editor a better starting point than a folder called “good TikToks.”
