Europe Is Building a Comparable Graduate-Tracking System. Your Satisfaction Mean Feeds Nothing Into It

Bottom line up front: Since the Council Recommendation of 20 November 2017 on tracking graduates, the European Union has been building something most course-evaluation systems were never designed to feed: a comparable, cross-border, longitudinal picture of what graduates do after they leave. The EUROGRADUATE survey is the instrument. A programme-level satisfaction mean, collected once at the end of a module on a five-point scale, produces no data that connects to this system at all. The problem is not that your evaluation is bad; it is that it answers a different, smaller question than the one Europe is now asking.

What EUROGRADUATE actually is

The 2017 Council Recommendation asked Member States to develop comprehensive systems for tracking tertiary and VET graduates at national level, and to improve the availability of comparable data at EU level. To test whether that was feasible, the European Commission funded the EUROGRADUATE pilot survey, first run across eight countries — Austria, Czechia, Croatia, Germany, Greece, Lithuania, Malta and Norway. The follow-on EUROGRADUATE 2022 expanded to seventeen countries. Its purpose, in the European Parliament's own briefing on developing graduate tracking at European level, is to lay the foundation for a sustainable, Europe-wide graduate survey covering labour-market transitions, mobility across Europe and graduates' wider role in society.

The keyword is comparable. The entire point is that a data point from a graduate in Malta can be read alongside one from a graduate in Norway. That ambition imposes requirements — on constructs, on wording, on scaling — that a locally-invented satisfaction questionnaire almost never meets.

Why your instrument produces nothing comparable

Three failures block a typical end-of-term evaluation from feeding a comparable system.

Construct. Satisfaction is not an outcome. EUROGRADUATE is interested in skills use, skills match, further learning, mobility and civic participation. "Overall I was satisfied with this course: 4.1/5" does not map onto any of those constructs. It is not a weaker measure of the same thing; it is a measure of a different thing.

Comparability. Even where two institutions ask about the "same" concept, a raw mean is not comparable unless the measure behaves the same way in both populations — what psychometricians call measurement invariance. Response styles, reference frames and translation differences mean a 4.1 in one system and a 4.1 in another are not interchangeable. This is precisely the wall the OECD's AHELO project hit when it tried to measure learning outcomes across countries: comparability is hard-won, not assumed.

Language. A European system is multilingual by construction. A satisfaction item translated ad hoc, without cognitive testing in each language, drifts in meaning — and the drift is invisible in the final average.

The result is that institutions sit on a rich longitudinal quality record — years of student feedback — that is structurally disconnected from the outcome infrastructure their own governments are now committing to.

But isn't this what tracer studies and destination data already do? The counterargument

The strongest objection is that graduate tracking is a destination activity — you survey alumni months or years out, as a graduate tracer study does — and it has little to do with in-course evaluation, so course evaluation is simply the wrong tool and no one should expect it to feed EUROGRADUATE.

That is half right, and the half that is right matters. Course evaluation cannot measure employment outcomes; those are lagged, and the lag is exactly why outcome data arrives too late to fix a live course. Destination surveys and in-course evaluation are different instruments doing different jobs.

But the objection misses the connective tissue. A graduate-tracking system that only records where people ended up cannot tell an institution why. To learn from EUROGRADUATE-style data, you need in-programme evidence about the proximal mechanisms — skills actually developed, alignment to intended learning outcomes, the quality of the experience — recorded in a form that can later be joined to destination data on the same cohorts. If your in-course instrument produces only an un-anchored satisfaction average, that join is impossible: you can see the outcome and the input, but nothing links them. The task is not to make course evaluation into a tracer study; it is to make it produce evidence that connects to one.

What comparable-ready evaluation looks like

Feeding a system like EUROGRADUATE means evaluation that: measures constructs (skills developed, outcome alignment, experience quality) rather than global satisfaction; is worded and scaled consistently enough to support comparison across programmes and languages; captures qualitative signal that survives aggregation; and is structured so cohorts can be followed and later linked to destination data.

Koji for Education is built for this shift. Its AI-moderated conversational interviews probe what students can now do and how the course produced it, not just how they felt — the proximal, mechanism-level evidence a destination survey cannot supply. Six structured question types (open_ended, scale, single_choice, multiple_choice, ranking, yes_no) let you standardise constructs across programmes instead of reinventing a form per department, and standardised, bias-aware AI moderation keeps wording and probing consistent across cohorts and languages, which is the precondition for any comparison. Automatic thematic analysis turns open-text into structured themes that aggregate without being flattened to a number, and programme- and institution-level reporting produces cohort-linkable records you can later read against national tracking data. It improves comparability and connective evidence; it does not turn a course survey into an official graduate register, and it does not eliminate the hard measurement-invariance problems — it just stops you from making them worse with an un-anchored mean.

The same conversational engine powers the main Koji platform, which institutional-research teams use for employer and labour-market research — the other side of the graduate-outcome loop.

Where this connects

This piece sits alongside the wider argument that course evaluation should measure what produces employability rather than employability itself, and the cross-border evidence bar set by the European Degree label. The common thread: Europe is moving toward comparable, outcomes-based, cross-border evidence, and the satisfaction mean is the one artefact that travels least well.

The takeaway

EUROGRADUATE is not a survey you fill in; it is a signal about where quality evidence is heading — comparable, outcomes-oriented, multilingual, longitudinal. Institutions that keep collecting un-anchored satisfaction averages will find they have decades of feedback that connects to none of it. The fix is not more surveys; it is evaluation designed, from the question up, to produce evidence that can travel.

Frequently asked questions

What is EUROGRADUATE? EUROGRADUATE is a European Commission-supported graduate survey developed after the 2017 Council Recommendation on tracking graduates. Its pilot ran in eight countries and the 2022 wave in seventeen, with the aim of building sustainable, comparable, Europe-wide data on graduates' labour-market transitions, mobility and civic participation.

Can course evaluation feed a graduate-tracking system? Not directly, and it should not try to replace destination surveys. But in-course evaluation can and should produce proximal, construct-based evidence — skills developed, outcome alignment, experience quality — recorded in a cohort-linkable form so it can later be read against destination data. An un-anchored satisfaction mean cannot be linked in this way.

Why is a satisfaction mean not comparable across institutions? Because comparability requires measurement invariance: the measure must behave the same way in each population and language. Response styles, reference frames and translation drift mean a 4.1 in one system is not interchangeable with a 4.1 in another. Global satisfaction also measures a different construct from the skills and outcomes EUROGRADUATE tracks.

Is this the same as a graduate tracer study? No. Tracer studies are destination surveys run on alumni months or years after graduation. This argument is about making in-course evaluation produce mechanism-level evidence that connects to such destination data, so institutions can learn why outcomes occurred, not just what they were.

How does Koji make evaluation more comparable-ready? Koji uses standardised, bias-aware AI moderation and structured question types to measure constructs consistently across programmes and languages, thematically analyses open text without flattening it, and produces cohort-linkable programme-level records. It improves comparability and connective evidence; it does not eliminate measurement-invariance challenges or act as an official register.