How the Career Atlas counts

What we count

• Skill fingerprint. For each role, the share of résumés in that role that list a skill (prevalence), and how much more common the skill is in that role than across all résumés (lift). Skills are ranked by distinctiveness, not popularity.
• Career moves. Where people go next and where they come from, read from dated work history inside the same résumé. Each move shows N, the count of real people who made it.
• Skill pairings. Which skills co-occur on the same résumés, ranked by co-occurrence strength, each pairing with its N.

Where the data comes from

The corpus is 102,822real US résumés assembled from public, research-grade résumé datasets in 2026. It is used only in aggregate: we publish counts and shares, never an individual résumé, and the Atlas carries no names, no contact details, and no personal data. Contact information in Futureful's job-matching product comes only from people who sign up themselves; never from this corpus.

What we refuse to do

• No salary. Résumés do not carry pay, so we hold no compensation data and never print a dollar figure on an Atlas page.
• No fabrication. When the evidence is thin, the page says so plainly. An honest empty state always beats a guess.
• No thin pages in search. A role or skill backed by fewer than 10 résumés is marked noindex, so search engines never see a page we would not stand behind.
• No individual exposure. Aggregation with support counts (N) is the only lens; small groups stay unpublished.

How to read an Atlas number

Every statistic carries N, the count of real résumés behind it. A skill shown as "62% · N=1,204" means 1,204 résumés for that role list the skill, and that is 62% of them. When you see "not enough data yet", that is the truth, not a placeholder. Corpus assembled July 2026; pages refresh daily as the Atlas grows.