The most common reply to the Russell Group piece was a fair one: if the label is not the instrument, what is? Rankings are the obvious answer, and they are useful. But only some of them, read in a particular way, and never on their own.
Institutional rankings answer the wrong question
A university's overall position is an average across everything it does. That average tells you almost nothing about the department you would actually join.
Glasgow published its 2026 subject results in March. Life Sciences and Medicine came 41st in the world. Natural Sciences came 149th. Same university, same year, same funding, a gap of more than a hundred places depending on which building you walk into.
The overall table would have given you one number for Glasgow. The subject table gives you the number that applies to you.
QS ranks 55 narrow subjects across 1,912 institutions. Times Higher Education recalibrates its eighteen indicators for each subject, giving less weight to citations in arts and humanities and more to industry collaboration in engineering. Both take the trouble to separate disciplines. Applicants often do not.
What the surveys are actually counting
In the QS World University Rankings, academic reputation carries thirty per cent and employer reputation fifteen. Nearly half the score is opinion collected by survey, from around 151,000 academics and 101,000 employers.
That is not worthless. Reputation shapes who reads your CV. But it lags reality by years, favours large and old institutions, and rewards visibility as much as quality. A department that transformed itself five years ago is still carrying the reputation it had a decade ago.
In some fields the surveys are the whole ranking. QS notes that in subjects like art and design, too few papers are published for citation data to be statistically meaningful, so the table rests on the academic and employer surveys alone. Worth knowing before you treat that number as measurement.
The data almost nobody opens
In the UK there is an official source that sits outside the league tables entirely. Discover Uni is owned and operated by the four higher education regulators: the Office for Students in England, the Scottish Funding Council, Medr in Wales, and the Department for the Economy in Northern Ireland.
It publishes at course level, not institution level, and it lets you compare up to seven courses side by side. Three datasets sit behind it.
- The National Student Survey. Final-year students rating teaching, assessment, academic support, organisation and learning resources. The 2025 round drew just under 357,000 responses, a 71.5 per cent response rate.
- Graduate Outcomes. What graduates are doing fifteen months after finishing, by course.
- Longitudinal Education Outcomes. Earnings at one, three and five years, drawn from tax records rather than a survey.
LEO is the strongest of the three because nobody opted into it. It comes from PAYE records, so it does not depend on who chose to answer.
Discover Uni gives its own warning about reading these figures, and it is the right one: only act on significant differences. Small gaps between courses are noise, not signal.
Where each source is useful
| Question | Where to look |
|---|---|
| Is this department strong in my field? | QS or THE subject tables |
| What do students on this course say? | National Student Survey, via Discover Uni |
| What happens after graduation? | Graduate Outcomes, by course |
| What do graduates earn later? | LEO, at one, three and five years |
| What will I actually study? | The programme page: modules, methods, placement |
Sources: QS and Times Higher Education methodology pages; Discover Uni, Office for Students; read August 2026.
The order matters more than the sources
Most applicants start with a ranking and work backwards to a reason. The useful sequence runs the other way.
Decide the direction first: what you want to be doing three years after graduating. Then find the subject table for that field and read it as a shortlist generator, not a verdict. Then take each candidate course to Discover Uni and check what students and graduates report. Then, last, open the programme page and read the modules.
That final step is the one people skip, and it is the one that decides your two years. Two courses with the same title at similarly ranked universities can differ completely in what they teach, whether there is a placement, who supervises the dissertation, and which methods you leave knowing.
What none of this measures
No table tells you whether a department is a good place to be. Supervision quality, how a cohort works together, whether staff answer email, whether the city suits you. These come from conversations: current students, recent graduates, the course convenor if they will speak to you.
A ranking narrows twenty options to five. It cannot choose among the five. That part is still judgement, and it should be.
Where I come into this
I do not build shortlists from rankings. I build them from the direction, then use the tables to test whether the list is realistic.
The work I do most often is the unglamorous kind: opening Discover Uni next to a programme page and asking why a course with strong graduate outcomes has poor student satisfaction, or why a highly ranked department has no placement route in a field where placements decide the first job.
Sometimes the answer is that the ranking is right and the course is excellent. Often the answer is that the number was measuring something the applicant was not asking about.
Is it the writing, or the decision?
We start with a short form. Six questions, two minutes. Which of the two needs work is usually clear on a first read.