What a skill-gap actually is

"Skill-gap analysis" usually means: take a target role, take a person, diff the skill lists, return what's missing. That diff is almost useless. It's long, it's generic, and it treats every missing skill as equally worth acquiring.

A more useful definition: a skill-gap is the ordered set of skills where acquiring each one measurably changes the person's odds of the outcome they want. Two claims are doing work there.

Ordered. Most of the value is in the first two or three items. A ranking that puts "system design" above "learn a second cloud provider" for a given profile is worth more than a complete unranked list. The ranking is the product; the list is a commodity.

Outcome-linked. "Missing relative to a job description" and "missing relative to people who actually got this job from a similar starting point" are different diffs, and the second one is the one that matters. Job descriptions are wish lists. Hiring outcomes are ground truth where you can get them.

This reframing has consequences for how you build the thing:

• You need outcome data (who moved from state A to state B), not just a skill taxonomy.
• Cold-start profiles (a first-year with no history) will get a weak, wide read, and the honest move is to show that uncertainty rather than hide it.
• The output should be "these three things, in this order, and here is why" rather than a score or a syllabus.

Aggregate patterns across the pilot cohorts, where these gaps recur and how they cluster, are in Aggregate skill-gap patterns across 1,500+ students.