What I Learned Publishing 15 TikTok Augmented Reality Effects in 1 Week
How Are Users Engaging With AR on TikTok?
Augmented Reality Seems Popular on TikTok
As someone who is passionate about the AR/VR industry, it’s hard to not notice the high levels of exposure that Augmented Reality Effects or “filters” have on TikTok. I was curious about how much engagement these AR experiences are garnering, so I decided to create and publish some “Effects” on TikTok to see for myself.
What I Built
I built and published 15 AR Effects on TikTok (Profile Page) over the course of one week:
- 2 Beauty Effects
- 2 Randomizers
- 3 “3D Mesh Anchored to Body” experiences
- 8 “Tableaus” of 2D layered scenes with slight movement
All AR Effects for TikTok are built inside of the TikTok-owned software “Effect House” (LINK). Here’s a screenshot of what one of these projects looks like in Effect House:
As you can see, it’s very similar to other software such as Snap’s Lens Studio (LINK) or Meta’s Spark Studio (LINK). By importing assets and combining built in tools for segmentation with visual scripting, a developer is able to create a virtually limitless amount of AR content, and easily publish directly to TikTok.
The Results
Here is the performance data after a month on platform (the last 5 entries were added two weeks ago at time of writing as additional AR Effects):
The Learnings
Impact Fades Fast
Daily Plays & Posts peak around 2–3 days after publication, and then crash. This pattern played out consistently across the different types of experiences, and across vastly different levels of individual Effect success. It’s hard to say if this is due to deliberate behavior of the TikTok algorithm or just a result of a consistent stream of new AR Effects pushing out older Effects, but notable of just how short the life spans are.
Plays Outweigh Posts / There’s More Consumer Engagement Than What is Publicly Visible
While many people might try out an AR Effect, only a fraction of those users will end up making posts using the effects. What’s telling about this is that users are interest in AR and are seeking out and trying different AR Effects, even if it’s not publicly visible behavior.
Randomizers Have High Play-to-Post Ratios (Bad), While Tableaus/Scenes Have Lower Play-to-Post Ratios (Good)
The chart above shows Plays per post (ex: for every 120 users who played the large yellow randomizer, only 1 person posted using the effect). For the two randomizers (large yellow column and large gray column), they had the highest play-to-post ratios, which means after trying out the Effect, more people decided not to actually post a video using the effect. However, Tableau based scenes that place a user in an environment had much better performance and have the lowest Play-to-Post ratios.
Takeaway
My main takeaways from this project are that users are seeking out AR content on social platforms, but not all types of AR content are equal in terms of generating posts. In this very limited case study, experiences that placed users in the center of Tableaus or scenes (such as background and foreground elements that make a user look like they’re sitting in the driver seat of a car) seem to perform higher than randomizers or beauty focused effects. I think future experiments looking at other platforms and continuing to explore with different types of AR content would be valuable to further identify what AR content is resonating with consumers.
Thank you for reading, and I hope these results can be helpful to you as you engage on your own path in the AR/VR community!
