A points-based migration system scores an applicant on characteristics correlated with national labour-market outcomes — age, qualification level, a language test band, prior in-country work — and ranks them against an intake quota. It does not measure job-specific capability. The score decides who may enter a labour market, not who can do a particular job.
The distinction matters more every year, because these grids are being rewritten in public. Canada published a discussion paper in April 2026 proposing to strip several factors out of its Comprehensive Ranking System, Australia's points test has been under review since 2023, and Germany's Opportunity Card runs on a six-point threshold introduced in 2024. If you hire across borders, the instrument that filtered your candidate pool is changing — and it was never built to answer your question.
What a points test is actually scoring
Three design features explain most of the confusion.
The criterion is not job performance. These systems are validated — where they are validated at all — against earnings, employment and fiscal contribution across a whole admitted cohort. Canada's figures show over 94% employment and median earnings of roughly CAD 58,000–80,000 one to five years after admission for Federal High Skilled principal applicants admitted 2015–2022. That is a real result, and a population-level one. It says nothing about whether a given applicant can run your month-end close.
The cut is set by quota, not by a standard. In a ranked pool, the pass mark is wherever the draw happens to land that round. A criterion-referenced cut score is derived from the performance a job requires; a quota-driven cut is derived from how many places are left. Both are legitimate, but only one of them tells you something about competence.
Single factors can swamp the grid. In Canada's CRS, age is worth up to 110 points, education up to 150, first official language up to 136, Canadian work experience up to 80 and skills transferability up to 100 — while a provincial nomination alone is worth 600. Australia's migration system review found bonus points were often "the primary determinants of individual rankings". A composite in which one administrative fact outweighs the whole human-capital section is not ranking people on human capital.
Three live designs, three different questions
Canada — compensatory ranking against a pool. Everything trades off against everything else, and candidates are ranked continuously. Job-offer points, previously 200 for senior managers and 50 for others, were temporarily removed in March 2025.
Germany — a low fixed threshold. The Opportunity Card needs a qualification recognised in its country of origin, German at A1 or English at B2, and at least six points: four for partial recognition of the qualification, three for five years' relevant experience within seven, three for German at B2 or above, two for being under 35, one each for a shortage occupation, prior German residence or a partner with skilled-worker potential. Applicants must also show roughly €1,091 a month in funds (2026).
United Kingdom — hurdles plus tradeables. The Skilled Worker route requires 70 points: a mandatory, non-negotiable block, with the remainder tradeable across salary, shortage status and qualification. Structurally this is a multiple-hurdle design wearing points-system clothing, and its thresholds have moved repeatedly.
Two candidates with similar scores in different systems have therefore passed tests of different constructs. The scores are not interchangeable with each other, or with anything on your scorecard.
Where the evidence says the weights were wrong
This is a rare public example of a selection instrument being audited against outcome data.
Canada's 2026 consultation paper proposes removing French-proficiency points, study-in-Canada points, sibling-in-Canada points and the spousal grid, describing the last three as relatively weak predictors. It reports that 76% of Canadian-study points awarded between 2023 and 2025 went to one- and two-year college credentials associated with slightly lower earnings; that language gains are modest at CLB 6 (about 8% more than CLB 4–5) and accelerate from CLB 7; and that candidates arriving without Canadian work experience tend to close the earnings gap within three to four years. It also notes that two of the strongest known predictors — pre-landing in-country earnings and a genuine job offer — are not in the grid at all.
The Grattan Institute's submission to Australia's points test review ran regressions on linked administrative data and found a similar pattern.
Factors that predicted earnings strongly
- English spoken "very well": about 2.5 times the earnings of "not well", and 1.9 times "well"
- A PhD: about 1.7 times a school-leaver; a bachelor's or master's about 1.4 times
- Nominating a skill level 1 occupation: about 1.6 times skill levels 3–5
- Prior Australian income of AUD 70,000–90,000: nearly double those under AUD 20,000
- A prior Temporary Skill Shortage visa: about 20% higher predicted earnings
Factors that did not
- Studying at an Australian university or VET provider, which "does not boost the earnings of permanent migrants"
- Studying in regional Australia, where the submission found no evidence of a long-term earnings benefit
- The professional year, which showed no earnings benefit
- A STEM postgraduate premium over other postgraduates
- A community language qualification
The submission also argues the age weighting is badly calibrated — a 25-year-old has roughly 14 more working years than a 39-year-old, but the test separated them by about five points — and that the design "rewards the quantity of work experience, rather than the quality". It notes the 2006 review behind Australia's 2010–11 settings relied on outcomes measured only six and 18 months after arrival: a criterion measured too early, then used to fix weights for a decade.
The number that should change how you screen
Across Australian linked data, roughly half of employed points-tested migrants were working in their nominated occupation one year after visa grant, falling to about 40% at 15 years. The occupation code on the visa is a statement of intent, not of work performed.
The European picture is consistent. In 2023, Eurostat put over-qualification among employed tertiary-educated non-EU citizens at 39.4%, against 31.3% for other EU citizens and 20.8% for nationals. Roughly two in five degree-holding non-EU workers are in jobs below their qualification level — so for a large share of people, the qualification signal that earned the points is not being used.
What to measure yourself
- Record the visa score as an eligibility fact, not as evidence. It belongs in the compliance file, not the scoring rubric. Double-counting it quietly reweights your process toward age and credentials.
- Assess the job, not the nominated occupation code. A work sample or AI-graded scenario built from the actual tasks tells you what the ANZSCO or NOC code cannot.
- Do not treat a migration language band as a job-communication measure. IELTS and CLB bands index general proficiency. If the role turns on explaining a billing error to an irritated customer, score that task.
- Verify the credential separately from the points claim. Recognition establishes that a qualification is comparable to a domestic one — a statement about the document, not about current performance.
- Hold the instrument constant across routes. Same assessment, same cut score, same administration conditions for domestic and visa candidates. AssessAll's AI proctoring with integrity bands exists so remote and in-person candidates can be compared on the same terms rather than on who happened to sit in a test centre.
- Make the evidence portable and dated. A Skill Passport that records what was assessed, at what level, when and under what conditions survives the candidate changing employer, visa class or country — which is more than a points score does.
- Check validity and adverse impact by route, not only overall. Pooled statistics hide a cut score that behaves differently for one cohort. The Uniform Guidelines and the Standards for Educational and Psychological Testing both expect evidence supporting the use you are actually making of a score.
When a points score is the right instrument
For a government allocating a fixed number of places among millions of applicants it cannot individually test, a transparent weighted composite is a reasonable instrument: the rules are published, administration is cheap, decisions are reviewable, and cohort-level outcomes broadly justify it. The Canadian and Australian reviews are not evidence that points systems fail — they show a system being recalibrated against better outcome data, which is what should happen. The error is in borrowing the output. A score built to forecast average earnings across a cohort over a decade was never designed to forecast one person's performance in one role next quarter.
A visa score tells you a candidate is admissible and that people resembling them tend to do well in that economy. Whether this person can do this job is a measurement you still have to take yourself.