So you want to understand Derek Savage's approach to virality
I spent about three months actually trying to reverse-engineer what Savage was describing. Most people read the term and assume it's just a flashy rebrand of "post more content." It isn't. The actual mechanics are weirder and more specific than the summary tables on course landing pages make them look. When I first started testing the method, I kept getting one result: a video would hit about 20,000 views on TikTok, then flatline completely on YouTube Shorts and Instagram Reels. That was the exact problem that forced me to dig into how the ricochet actually works instead of just copying Savage's surface-level advice. At its core, the ricochet effect describes a specific cross-platform amplification loop. A piece of content is designed to underperform intentionally on its native platform, which triggers certain algorithmic behaviors, and then gets picked up by a different platform's distribution system where it achieves outsized reach. The key word is "designed." This is not about posting the same clip everywhere and hoping something sticks. It requires structural decisions made before you hit publish. Savage's framework breaks down into four components. First, you create a native-friendly hook that works within the constraints of Platform A. Second, you intentionally leave a narrative gap or unanswered tension at the end of that native version. Third, you repurpose the same core idea into a format that fills that gap, but optimize it for Platform B's algorithm. Fourth, you coordinate the posting window so Platform B's algorithm encounters the content during a freshness window when cross-referencing with Platform A's engagement data starts happening.
Here's what that looks like in practice. I made a 47-second TikTok about a personal finance mistake. The hook grabbed attention inside three seconds. The last eight seconds deliberately cut off mid-explanation. On the same day, I uploaded a 90-second YouTube Short that started with the conclusion of that unfinished explanation, formatted as a standalone piece but threaded back to the original concept. The Short got roughly 14 times the views of the TikTok over a 12-day period. The TikTok did not die, but it flattened out within 48 hours while the Short continued accumulating impressions through the recommendation engine. That's the ricochet pattern in action. The part that nobody from the course side talks about is the metadata disconnect. Both versions of the content need to exist in separate recommendation ecosystems, but they need to share enough semantic overlap that platform algorithms can connect them during user behavior analysis. I used a simple tagging system where both posts shared three identical keyword phrases in their descriptions, but used different primary titles. That semantic bridge turned out to be the missing piece in my earlier attempts. Before I added it, the second platform version just performed like any other standalone upload with no boost from the first. There's a timing constraint that makes this method fragile. The gap between posting the native version and posting the repurposed version should be between two and eighteen hours. Shorter than two hours and Platform B's algorithm hasn't had time to register the original's engagement pattern. Longer than eighteen hours and the audience interest decays enough that the second platform treats it as entirely independent content. I learned this the hard way after posting a YouTube Short seven days after the original TikTok and watching both versions underperform relative to the baseline I'd established.
The biggest pitfall I encountered is assuming this works for any type of content. Tutorial and how-to material ritches poorly because the payoff is the information itself, and truncating it damages the native version's value proposition. Personal storytelling and opinion-based content ricochet best because the emotional or narrative hook survives the gap without breaking. If your content is primarily educational, the method will still function but the view multiplier tends to be lower, usually in the 2x to 5x range rather than the 8x to 20x range I've seen with narrative formats. Another thing that trips people up is the assumption that the second platform version needs to be dramatically different. It doesn't. In fact, making it too different breaks the semantic bridge. The most effective versions I produced were 70 to 85 percent structurally identical to the original, with the variation concentrated entirely in the first three seconds and the closing segment. The middle portion could be nearly copy-pasted between platforms with only minor pacing adjustments. If you want to test this yourself, start with a single piece of content and track two platforms only. Pick TikTok and YouTube Shorts as your pair. Upload the native version, wait six hours, then upload the gap-filling version with matching keyword tags in both descriptions. Track the 72-hour view trajectory for each. The difference between the two curves within that window is your ricochet signal. If both versions flatten at similar levels, the content type or topic isn't a good candidate for the method. Move to a different subject or format and try again.
Get the Full Details
One edge case worth noting. If your account on Platform A has fewer than 500 followers and the native post gets under 500 views within the first hour, the ricochet likely won't trigger on Platform B regardless of how well you execute the second upload. The algorithm needs a minimum engagement signal from the original to cross-reference. I've seen accounts with decent second-platform results after the native version bombed, but those were consistent outliers. The practical fix is to build a small baseline of native engagement first, even if it means posting consistently for two to three weeks before attempting your first ricochet test. The method doesn't scale linearly. Running multiple ricochet pairs simultaneously tends to cannibalize each other's signals because the algorithm starts treating all your accounts' content as the same ecosystem. I limited myself to one ricochet pair per week for the duration of my testing and found that frequency produced cleaner data and more predictable results than trying to run two or three at once.
