Methodology

The Career Shock Score measures how much your work is likely to be transformed — tasks changing hands, tools replacing steps, output expectations shifting. It is not a probability that you lose your job.

Scores, insights, statistics and career matches are computed by fixed formulas and lookup tables in your browser. No language model, no external inference, no tokens — before or after you verify your email.

0–34 low, 35–54 moderate, 55–74 high, 75–100 severe transformation.

Uses only answers collected before the reveal: relevant real AI use cases (45%), workflow depth (45%) and the resilience of occupation-derived human skills (10%). Tool count and age are never inputs.

Skill overlap 30%, preference fit 15%, forward opportunity 15%, AI-era resilience 15%, retraining feasibility 10%, experience transferability 10%, education fit 5%. Hard filters remove roles blocked by credentials or beyond your stated retraining limit. Distance weights missing skills 45%, credentials 25%, experience gap 15% and retraining willingness 15%.

Rises with the number of tasks we could classify and whether your occupation is in the seed set. It never reaches certainty.

This is a beta with seeded, directional estimates. It does not forecast salaries, hiring volumes or the fate of any individual role, and it cannot see your employer, your market or your manager.

Every statistic shown during the assessment comes from a fixed, versioned registry of approved, dated public sources — labour statistics agencies, peer-reviewed studies, regulators and named industry surveys. Each one carries its scope, geography and year, and we link to the original. We never turn an exposure or forecast figure into a job-loss claim.

The score combines two axes. Market pressure (60%) describes your occupation: occupational exposure, substitution pressure, demand outlook risk, client self-service pressure, adoption velocity and entry-level compression, weighted 20/20/20/15/15/10. Your task pressure (40%) describes your own mix: task exposure 40%, automation tendency 25%, capability pace 20% and the inverse of human necessity 15%, with your two primary tasks weighted 1.75x. An industry adjustment of at most plus or minus 8 points is applied on top and shown in your result. Your age and your AI skill are never inputs. The score is an index of transformation, not a probability of losing your job.

Every occupation resolves to a market-pressure profile. Some roles have an occupation-specific profile grounded in named sources; the rest inherit their job family's baseline, which is marked in your result and carries lower confidence. We do not invent occupation-level precision we cannot support.

Career Shock is about the market. AI Readiness is about you. Being good with AI never lowers Career Shock, because your skill does not change the pressure on the occupation — it changes what you can do about it. High shock with high readiness means a fast-changing field you are positioned to lead in; high shock with low readiness means urgent action; low shock with high readiness is an advantage; low shock with low readiness is stable for now, but thinly prepared.

How you work adjusts the score by at most -4 to +6 points, never more. Being an employee, a freelancer or a business owner does not change the pressure on your occupation itself — the base score stays identical. The adjustment uses only your explicit work setup: income structure for independent work, or the organisation's current AI stage for employees. Both the base score and the adjustment are shown to you.

Readiness V3 is built only from information collected before the reveal: relevant real AI use cases (45%, capped so listing everything cannot inflate the score), workflow depth (45%) and occupation-derived human strength (10%). Tool categories are kept for context but never scored. Verification habits, learning time, change appetite, education and priorities are not assumed or defaulted.

The minimum number of tasks depends on how many are actually defined for your occupation. Well-covered roles ask for at least three, medium coverage two, and thin coverage one — nobody is forced to tick boxes that do not describe their work. You may select up to seven, and the two you mark as primary carry extra weight.

Every occupation belongs to a domain, and every technology card is tagged with the domains it belongs to. A card from another domain can never be shown to you unless it is explicitly marked as domain-neutral. An event planner will not be shown a drive-through or hotel front-desk example; if nothing relevant exists, we show nothing rather than filler.

Some insights describe a specific technology and the exact task it performs. Each one is labelled with its honest maturity: already in use, pilot, announced, research or forecast. 'Already in use' means documented production use — a vendor announcement or a research result is never presented as widespread deployment.

Occupation matching, scoring, insights, archetypes, recommendations and share images are all computed deterministically on your device or by plain code on our server. No language model is called at any point — before or after you verify your email.

The statistics shown during the assessment come from a static, versioned registry. Each entry carries its source, publication date, geography, sample scope and evidence type, and only reviewed, unexpired entries can appear. Forecasts are always labelled as forecasts, and exposure is never presented as job loss.