Gap score is a real 0-100 score, calculated per keyword, that answers one question: does search demand for this keyword outpace how well the existing top-ranking videos already satisfy viewers?
Quick answer
Gap score is a 0-100 score, calculated per YouTube keyword, measuring whether search demand outpaces how well existing top-ranking videos already satisfy viewers -- derived from like-to-view ratio, comment-mined complaint patterns, and video recency, not from search volume.
Gap score is derived from three real inputs: the like-to-view ratio on top-ranking videos for that keyword, the frequency of specific tracked complaint patterns mined from real comments, and how recently the best-ranking video was made. A high score means real, checkable room for a better video to win; a low score means the keyword is already well-served. It's calculated per keyword, not once per niche, since the real opportunity concentrates in specific sub-niches and keyword clusters rather than applying evenly across a whole topic area.
MASSIVE gap (70+): existing videos are failing viewers badly -- the clearest opening
HIGH gap (50-70): clear room for a meaningfully better video, not a total vacuum
SATURATED (under 30): the keyword is already well-served, and high competition usually follows for a reason
Recalculated against whatever videos currently rank, so a keyword that was SATURATED a year ago can show a real gap again if the top-ranking content has gone stale
Does a high search-volume keyword automatically have a good gap score?
No -- volume and gap score measure different things. A keyword can have enormous search volume and a low gap score if the existing top-ranking videos already satisfy viewers well; the gap only exists when demand and satisfaction genuinely disagree.
Is gap score the same across every niche?
The calculation method is identical everywhere, but what counts as a strong gap in practice varies by niche, since baseline like ratios and comment volume differ by content category.
Can gap score change over time for the same keyword?
Yes -- it's recalculated against whatever videos currently rank, so a keyword that looked saturated a year ago can show a real gap again if the top-ranking content has gone stale or the topic itself has changed (new tools, new rules, new pricing).
Is the comment-pattern scan an AI guess at sentiment?
No -- it's deterministic pattern matching against real fetched comments for specific tracked phrases ('didn't answer', 'outdated', 'too basic', and others). Claude's role is limited to suggesting a content angle from the real patterns found, not inventing the patterns themselves.