Content creators frequently describe a TikTok “second wave,” or “second push,” as a red-carpet style secret passage: The video will generate a small volume of total views, then the views might slow down, and there appears to be another sudden resale of a similar video to current videos of the same stage. This can seem like a chance occurrence, but it makes sense once you start viewing it as a normal mode of generating views as opposed to a vibe.
I see many musicians and smaller businesses panic and change their editing after the first hour, but it’s actually a better option to wait and read early signals. If you have good completion and saves of an initial post, but are stuck with reach, I suggest leaving the creative and just doing a controlled test (for example, similar to aTikTok promotion pack, and you may find a better audience through the new test.
TikTok does not offer precise scorecards for each video, but they have been fairly consistent through the various participant groups since they all have some common characteristics that determine when your video will be tested with other new audiences. This article discusses those commonalities, examples of what to expect to see when they perform well against each of them and recommendations on how to adjust your content should one or more of the areas become a drag on overall video performance.
What a “second push” really is
Distribution works similarly to movie test audiences. TikTok tests your video with a small but diverse audience; based on their reactions, TikTok will decide to increase distribution; and if there are good metrics to support the decision, TikTok will test again with a different audience later (this is referred to as a ‘second push.)’
A good way to illustrate this principle: recommendation systems utilize behavioral signals from how users have previously watched content more than just by the “like” button. The link below to the page discussing TikTok’s Algorithm shows watch time and completion percentage as two very important signals of how well a video does on the platform; hence, the reason a video may appear to have been “dead” with 2000 views until it locates true viewers that will watch it through to completion.
The signals that usually earn wave two
You don’t have to get the exact numbers from your measurements; what you want is an indication of how you perform relative to yourself and/or other videos of similar length. Generally speaking, one weak signal in a video can interfere with the strength of any other signals.
- Normally, completion rates by category will be as follows: under 40% completion is bad news for nearly all posts between 15-30 seconds long; 50-70% completion is not too bad; and over 70% is usually a positive response – particularly if they have a steep decline in the last portion of the video. Longer videos will generally have lower completion rates, so when evaluating completion rates make sure to compare apples with apples regarding similar lengths.
- Rewatch Behavior – A viewing event that occurs when something in the video has a lot of “density” (i.e., a reveal, punchline, beat drop, or quick tutorial). When this occurs, if the average watch time exceeds the length of the video, you have virtually printed a “show this again” ticket.
- Comment velocity(the number of comments rate) is more important than total number of comments. An example would be ten comments made within the first twenty minutes of a video being posted will generate a higher comment velocity than one hundred comments made across forty-eight hours. The earlier the received comments show interaction, the more likely the video has generated an engaging experience.
- Within the first one to two seconds of viewing, if a user is bailing out because they are “swiping off” of your video, TikTok will assume that your video was an incorrect match for them. These “swipe offs” will cause your video to suffer without being able to see through any engagement, because your likes will come from the remaining viewers who continued. Most users left without saying or giving an indication they were leaving.
One additional nuance of intent is that both saves and shares carry more weight than likes when working with tools that help us understand Long Tail Distribution. An Industry explainer, Loomly, states the user interaction signals and watch patterns that are driving the For You feed, can be seen and experienced by using these same markers to illustrate how a post with many saves can experience a plateau, before being retested later.
A diagnostic flow you can reuse
This is the trouble-solving tree I utilize with artists teams. It’s not glamorous, but it eliminates the “post consistently” hand-waving and brings you back to making edit decisions that are within your control.
- If there is a quick drop-off in the first 2 seconds of your video, fix your hook before doing anything else with the video.
- If the hook holds but completion is low: your pacing or structure is the issue.
- If the completion feels good but the rewatch is not hitting: then there is probably something within the loop or dense moment worth revisiting.
- To improve your comment rate when the watch metrics are good but your comments are very low, create clearer conversation prompts and/or create more polarizing angles.
- There are two reasons why growth could either stop dead or slow to a near halt; Either the topics you are targeting are too niche or the initial target audience match for them was incorrect.
To fix fast swipes (high swipe rates on Hook cut), you should start with the payoff, not the setup. For musicians, this may mean beginning with the line of the chorus that will have the greatest impact and then showing the behind-the-scenes footage. Similarly with Brand films; start with the result shot first; then tell them how it was done. Generally speaking, the majority of people get caught up in trying to create a cinematic feel when in fact TikTok tends to reward clarity more than ambiance.
To improve pacing (when your delivery is slow): eliminate silence on the audio track; tighten the breaks between lines; delete any throat-clearing sentences. If your video has a length of 25 seconds and your message occurs at 19 seconds then reduce it to a new play length of 16 seconds. Look also for visual cues that refresh every one to three seconds (jump cuts, problems changing the camera angle, updating on-screen text). If no change visually occurs, your audience will likely lose interest.
Captions fix errors (when the message doesn’t convey). Caption audio to further the story-meaningful captions. For example – “New song coming out,” example – “I have re-written my hook 12 times… here is version 12”. Captions are even more critical for music creators because people watch with low sound values.
When it comes to subject-specific fixing (where the material itself seems acceptable, but the area of dissemination is limited), you will want to alter the method’s presentation. I have worked with different teams, and one of those teams posted a “new EDM track” to the studio and got a moderate response., but I presented the same video (studio) clip from another #Angle as “how to make your telephone speakers sound like DJ turntables.” That video was 2nd pushed when people searched for video tutorials. TikTok has a large number of videos based on interest, and this is the traditional human way to define that interest.
Managing several content accounts can be highly challenging; however, if you’re using Promosoundgroup for your distribution, you can treat the distribution as a controlled experiment rather than a daily panic. The goal is to provide clean samples of interest in order to deliver your posts to the right viewers early enough in order to create momentum.
The 1-variable experiment template
A lot of times, teams will often pursue “ghosts” due in part to doing 5 different things at once; changing the hook, changing the length of post, changing caption styles, changing up their hashtags and posting times and then will say the algorithm is being “moody.” Don’t do this.
- Pick one weak metric (hook drop-off, completion, rewatch, comments).
- Change one variable only (hook line, first frame, cut length, caption style, CTA question).
- Hold everything else steady: same topic, similar length, same posting window, same format.
- Run 3 posts minimum for that variable before you decide anything.
- Write easy-to-understand log entries that show results such as “held better at 1s”, “more rewatches after looping”, and “comments doubled after asking for this (X)”.
Lets take a real-life case of a musician creating 3 content posts about the same song using a snippet. The first post has the song’s choruses captioned out loud and bold. The 2nd post would have the singer say something like “just a heads up, there’s a drop in this song.” (spoiler: people are more likely to swipe through) The final post would use the chorus as well as include a prompt by using 1 line with either “this reminds me of summer, or this reminds me of heartbreak.” By doing so, you will know if the changes in the first and last posts have been made, something other than your imagination.
The most significant winner by far – creating the habit of viewing the data as signposts instead of determinations about your ability as a creator. Although the way TikTok weighs information will change slightly yearly, the patterns of where to distribute content, when viewers are actively watching, and matching audience to your content remains the same.
A second push is almost never magic. It’s TikTok having enough evidence to decide to take another chance. Your role is to provide that evidence in a clear way: a hook that stops thumbs, pacing to get people to finish watching, a loop that gets them to watch again, and an angle that attracts the right audience.
