TEF and student experience: what the metrics don't capture

UK higher education's most consequential external assessment — the Teaching Excellence Framework — claims to measure something called the "student experience." The metrics it uses to do that are the National Student Survey, continuation rates, completion rates, and Graduate Outcomes survey responses fifteen months after a student leaves university.

This is a strange definition of student experience.

It's not that the metrics are wrong. They measure real things, well. But they measure outcomes, not the experience itself. NSS asks final-year students to score satisfaction across five themes. Continuation rates record who stayed and who left. Graduate Outcomes records salaries and employment status after the dust settles. The thing that produces those numbers — the actual texture of three years lived inside a university — sits in the gaps between them.

This article is about that gap. What TEF measures, where the measures fall short, and what universities can do about the rapidly closing window before the next round of assessments begins.

What TEF actually measures

The Teaching Excellence Framework is run by the Office for Students, the independent regulator for higher education in England. Its most recent assessment, TEF 2023, covered 228 universities, colleges, and alternative providers. Of those, 46 received Gold ratings, 100 Silver, and 29 Bronze; a further 53 were pending at the time of initial publication. The ratings sit on top of a regulatory baseline that all OfS-registered providers must meet regardless of TEF outcome.

For the first time in 2023, providers received not one rating but two: one for student experience and one for student outcomes, combined into an overall award. The Office for Students described this two-aspect structure as providing "richer information than looking at the overall ratings alone."

What feeds those two ratings is a familiar set of indicators. The student experience aspect uses five NSS themes: teaching on courses, assessment and feedback, academic support, learning resources, and student voice. The student outcomes aspect uses continuation rates (did students make it past year one?), completion rates (did they finish?), and progression — measured by the HESA Graduate Outcomes survey, which contacts every UK graduate roughly fifteen months after they complete their course.

Alongside these indicators, providers submit a fifteen-page narrative document. Students submit a separate document independently. The TEF panel — drawn from academic staff and student representatives — weighs the indicators against the submissions and assigns the rating. Indicators contribute no more than half the evidence of excellence; the rest is the narrative.

That last sentence matters more than it sounds. In a Gold/Silver/Bronze system where many institutions cluster close together on the numerical indicators, the narrative is the difference. Two universities with materially identical NSS and continuation scores can land in different bands depending on how well they evidence what's behind the numbers. The question, then, is what counts as good evidence.

The metrics' blind spots

Four observations about the TEF indicators are uncontroversial in the sector and worth stating plainly.

First, NSS is retrospective. It asks final-year undergraduates to score their experience in the months before they leave. By the time the data is collected, the student has finished. Whatever it captures about the experience is filtered through end-of-degree perspective: relief, exhaustion, nostalgia, future anxiety. It is also annual, so any institutional change introduced midway through a cohort's degree shows up in the data — if at all — years later.

Second, continuation and completion are outcomes, not explanations. Knowing that 91% of a cohort continued to year two tells you nothing about why 9% didn't. The students who left took the reasons with them. The students who stayed may have stayed for reasons that have nothing to do with anything the institution did or didn't do.

Third, Graduate Outcomes runs fifteen months after the student leaves. HESA's annual survey contacts approximately 700,000 graduates each year; the most recent published data had a 44% response rate. The fifteen-month gap is, the agency explains, intended to give graduates "a meaningful opportunity to progress in their post-graduation activities." This is a reasonable trade-off for the metric it produces. But it produces a graduate metric, not a student-experience metric. By the time it's measured, the student isn't a student.

Fourth — and this is the one that most directly motivates what follows — none of these capture belonging. Belonging is the term sector research has converged on for the lived sense of being part of a community, fitting in, having peers, knowing how to ask for help. There is now a substantial body of evidence linking belonging to all of the outcomes TEF actually measures. AdvanceHE's analysis frames it as "a crucial element for academic success, student retention, and overall wellbeing." The 2022 HEPI / Advance HE Student Academic Experience Survey found just 45% of students felt they belonged at their institution.

The connection runs from belonging to retention to continuation rates to TEF ratings. But it runs in that direction. The metric at the end of the chain — continuation — can tell you the chain broke somewhere. It cannot tell you where.

What the gap looks like in practice

Consider two cohorts at two universities.

Cohort A has good attendance, decent assessment marks, and a strong continuation rate. The cohort's NSS scores come in above benchmark. On the metrics that TEF measures, the institution looks healthy. But inside the cohort, there are students who have made no friends, who haven't been to a society event, who eat alone, and who plan to commute home every weekend until graduation. They will probably finish. They will probably even score the institution positively on the NSS. But the experience they had is not the experience the institution is trying to deliver, and the metrics will not flag it.

Cohort B looks identical on every TEF indicator. But its students are embedded — in societies, in sports clubs, in study groups, in the local community. The continuation rate is the same. The NSS scores are the same. The eventual Graduate Outcomes results will probably be similar. From outside, the two cohorts are indistinguishable.

This is the kind of distinction the TEF narrative submission is supposed to capture. In practice, most universities struggle to evidence it beyond survey extracts and anecdotes. The OfS itself acknowledged after TEF 2023 that the provider submissions varied widely in the depth of evidence offered for student experience claims. The panel's summary statements, published alongside ratings, give a window into how submissions were weighed.

The pattern is consistent: where two institutions are otherwise comparable on the indicators, the one with concrete, segmented, longitudinal evidence of how it delivers the student experience outperforms the one relying on satisfaction quotes. The gap between Silver and Gold often lives precisely in that evidence layer.

This matters more under the next TEF than it did under TEF 2023. From the proposed integrated quality system onwards, a Bronze rating will explicitly mean only that a provider meets the minimum quality baseline — not that it demonstrates excellence. Under proposals out for consultation in late 2025, Bronze-rated providers may also face restrictions on student number growth. The cost of weak narrative evidence rises accordingly.

What good narrative evidence looks like

The TEF 2023 panel's published statements reveal what panel members rewarded. Three patterns recur.

Evidence is segmented by student characteristics. A claim that "students benefit from a strong sense of community" is weak. The same claim broken down by Access & Participation Plan target groups — commuter students, mature students, BAME students, students with disabilities, students from low-participation neighbourhoods — is much stronger. The panel cannot easily compare benchmark satisfaction scores across these groups (NSS data is benchmarked at the provider level), so segmented evidence about experience is genuinely additive.

Evidence is longitudinal. A single-year snapshot is easy to dismiss as an outlier. Three years of trend data on the same metric — engagement with society events broken down by year of study, or year-on-year change in proportion of first-years reporting close peer connections — speaks to whether institutional interventions are working. The TEF four-year cycle (and, under proposals, the new three-to-five-year rolling cycle) rewards institutions that have been collecting data continuously, not those that started collecting in the year before submission.

Evidence connects intervention to outcome. Panels respond to causal chains, not correlations. An institution that can show "we introduced peer mentoring in commuter cohorts in 2022, peer-mentoring participation rose to 62% of eligible students, continuation in that group rose 4 percentage points relative to the comparison group" has done the work. An institution that can show only the continuation improvement has not.

What none of this requires is a new metric. It requires the data the institution already generates — society memberships, event attendance, peer-connection participation, internal pulse-survey responses — to be captured systematically, demographically segmented, and linked back to formal student-record data. Most universities have most of this data. Few have it in a form that can survive a TEF panel's scrutiny.

This is the infrastructure gap that platforms like Reaction are starting to fill: capturing peer-connection and engagement data continuously, broken down by APP demographic categories, and exportable into the formats panels actually read. The broader point is that the gap can be filled by systematic data capture in whatever form an institution chooses — but it needs to be filled before it's needed, not after.

Where this goes next

In June 2025, the Office for Students confirmed publicly that the TEF will not run as a one-off exercise in 2027. The new approach — currently out for sector consultation — proposes a rolling assessment cycle where every registered provider is assessed every three to five years depending on its current rating. Gold-rated providers cycle every five years; Silver every four; Bronze every three. All providers will participate, including the smaller colleges that previously opted out.

Three things follow for institutions thinking about evidence.

First, the calendar advantage of preparing close to a submission deadline disappears. Under a rolling system, the next submission could be a year away or five years away depending on the current rating. The institutions that win are those that have been collecting the right evidence continuously.

Second, the proposals suggest a greater weight on direct student input — not just the student submission, but evidence of how students contribute to and benefit from the experience. Belonging, peer connection, and engagement-based evidence become more central, not less.

Third, the Bronze rating is being explicitly redefined as "meets the minimum quality requirements," with potential consequences for student recruitment caps. The distance between Bronze and Silver — historically a difficult call for panels assessing similar-looking institutions — becomes a high-stakes distance.

For Pro-VCs, Directors of Student Experience, and the teams that prepare TEF narrative submissions, the implication is clear. The infrastructure for evidencing student experience needs to exist before it's needed.

Closing

TEF's metrics are not going away. They were designed to be the comparable, benchmarkable, regulatory backbone of UK higher education quality assessment, and they will continue in that role through whatever the integrated system becomes.

The gap between what they measure and what students actually live will also continue.

Institutions that build the evidence layer above the metrics — peer connection data, belonging signals, segmented engagement records — will own the narrative space where TEF panels make the close calls. The ones that don't will keep submitting strong NSS scores and hoping the panel reads between the lines.

Pro-VCs we're speaking to are already thinking past TEF 2023 to whatever the integrated system becomes. The institutions that come out well will be the ones that started measuring what mattered before being told to.