Cracking the Algorithm: Lessons from Our Fallout Costume Reel Experiment
It all started with a fun idea – what if we could human-render the Fallout AI filter that was trending on social media? And would it be possible to make a video go viral by triggering certain engagement metrics in a coordinated, global effort?
Well, we tested it out and the answer is…no!
As a content creator in the food niche, I’ve had the opportunity to connect with Meta Content Experts to better understand their algorithm and the strategies that help push content to more users. While some of the below tips might not be new to content creators, I received a lot of great insight into best practices:
- Use trending audio (identified by the arrow ↗ beside songs)
- Lean into reels and post 3-5 times a week and keep this up consistently
- Schedule content during vacations, because missing a week of reels will cause the algorithm to drag behind when you come back
- Make carousel posts occasionally to engage audiences
- Interact with the community you build with polls, questions, and CTAs
What about the metrics? Which metrics matter?
You know we’re data and metrics nerds here at Fullintel. After all, it’s common knowledge that PR professionals regularly use data to inform their communications strategies! In order of observed impact (high to low), the following engagement metrics all contribute to pushing an Instagram Reel:
Shares – Users sending a reel to others indicates to the algorithm that the content was good enough that others should watch it.
Average View Time – All social platforms want their users to spend more time in the app, so higher average view times are great indicators of potential virality.
Saves – Users saving a reel tells the algorithm that the content was good enough that they will revisit it later.
Comments – User engagement via commenting is valuable because they’re spending time in that reel to write a comment, which also contributes to the average view time.
Likes – Lowest on the list, likes still indicate positive user engagement!
The Theory
To gauge whether a reel should be pushed to more users, the algorithm scores all the above metrics during the first couple of hours after publishing. If the video has a high engagement rate across the board – especially with shares and average view time – the video would have a much higher likelihood of trending and going viral, right?
The Experiment
We conducted the experiment, by coordinating with our entire company to engage with the Fallout Costume video within the first hour of publishing. To do so, we published the video on my Instagram account and set up an email and calendar reminder for all our colleagues and peers across multiple regions and time zones to engage with the video directly.
This resulted in hundreds of likes, comments, and shares within the first 30 minutes, as shown in the screenshots below:
With an engagement rate of 50%, and an average watch time of 35 seconds (of a 37 second long reel), that should trigger some level of virality, right?
Wrong. While the video performed decently well and reached a play count of 22.4k views (at the time of this blog’s publication), this cannot be considered viral.
Why? Don’t the metrics and high levels of engagement count?
While the coordinated efforts in boosting the video’s engagement did help push the reel to some new users, we strongly believe that this form of external sharing (as a link outside the Instagram platform) does not contribute to the algorithm’s benchmarking of video engagement.
A reel has to perform well and be driven by the organic push, meaning only the engagement metrics of the views it pushes within its own environment count. The evidence? Sending a link to a reel via email and engaging hundreds of our colleagues did not contribute to the benchmark metrics.
Conclusion
We’re confident in concluding that it is NOT possible to trigger virality with any external efforts. Everything has to be organic and within the Instagram (or TikTok) platforms!
However, it is worth noting that this is only part 1 of our experiment. The goal of this video was to mark whether we could push a costume trend video in May 2024, when there is almost no costume content trending. Part 2 will be re-posting this same video and concept in October 2024, when Halloween and costume content is much more dominant.
If you would like to check out our video that we had so much fun making, check it out here.
And for any inquiries on monitoring social data and metrics to inform content strategies, please feel free to reach out to me at anguyen@fullintel.com or book a consultation below!
