Here's the honest answer first: virality is a compound event, timing, mood, one early share landing with the right person, and no checklist controls it. What is predictable is whether a video is built to travel if it gets the chance, because the currencies that move reach are structural, and structure can be scored before posting.
The pre-flight is seven checks, one point each. Does the hook score three or better on the rubric? Is a question still open at the midpoint, or does the video answer itself early and bleed? Could the first two seconds survive losing another half-second? Does something change visually or vocally every 8 to 12 seconds? Can you name the specific person a viewer would DM this to, because everyone means no one? Is there a save reason, something worth returning to? And does it end at the payoff, with no outro? Six or seven: post. Four or five: one targeted fix. Under four: it's a draft.
When the score says fix, fix by removal, never by adding. The context paragraph, the second example, the warm-up sentence. The edit that travels is the one with nothing left to cut. And the pre-flight never overrides the calendar: a 5-of-7 posted today beats a 7-of-7 that made you miss three days.
Then the other half: the next morning, read three numbers only, 3-second retention, completion, and shares plus saves, and write one sentence in a verdict log. At thirty entries, patterns emerge. At ninety, you own a personal algorithm map no guru can sell you.
The full system is Companion 2 of our Yap Companions toolkit.
You can't predict lightning. You can build the rod.