What can SkillDrift actually prove today?
Three things, all of them measured inside the product rather than reported by users. First, that a readiness score changes over time for people who complete learning, because the score is recomputed from the same resume every time it is read. Second, that a completed course writes a certificate into the resume without the user doing anything, because that write is the product mechanic rather than a manual step. Third, that a job match score out of 100 moves when the underlying skills change, because the match is recomputed against the same job description.
All three are statements about the machinery. None of them is a statement about employment.
What does SkillDrift count as a result, and what does it deliberately not count?
We count things the product does: a readiness score recomputed, a named gap closed, a certificate written into a resume, a tailored version generated, a job match re-scored. We do not count applications sent, interviews attended, offers received or salary changes, because we do not observe any of them.
A user can be hired the week after they close a gap and we would never know, and a user can close every gap and not get hired. Anyone claiming a career platform has measured your employment outcome is either integrated with your employer or guessing.
Is any of this independently verified?
No. Every number on this page is measured by SkillDrift from SkillDrift’s own data, and none of it has been audited by a third party. That is a real limitation and it applies to every AI career product currently on the market, ours included.
The reason to state it rather than bury it is that a score you cannot interrogate is worth less than a score you can: our numbers are worth exactly as much as the method behind them, which is why the method is on this page.
What would actually prove it, and what would have to be true?
Outcome validation needs three things we do not have yet: a large enough cohort followed for long enough, a control group of comparable people not using the product, and a reliable signal of what happened to them afterwards. The honest position is that none of the three exists today.
What we can do in the meantime is publish the method for every number we quote, correct anything found wrong, and never present a company reported figure as an independent one.