The Transcript Problem: Why the Record That Lasts Longest Says the Least 

Blog Series · Academia Track · Part 6 

So far this track has looked at the evidence behind a single result. Part 1 showed that a scorecard is not evidence on its own. Part 2 asked whether two examiners would mark the same script the same way. Part 3 followed the paper trail behind one result. Part 4 tested what breaks at the scale of ten thousand screens. Part 5 asked what an institution can promise when someone wants to know how a result was reached. 

Most of the time, though, nobody asks. The appeal window closes, the result is settled, and the evidence goes into storage. What leaves the institution is the transcript: course titles, credits and grades. 

That page is read by employers, admissions panels and licensing bodies for decades. Part 6 asks what it actually tells them, and whether it can ever say what a learner is able to do. 

The result outlives its evidence 

By the time a grade reaches a transcript, almost everything that made it meaningful has been left behind. The marks against each criterion, the examiner’s comments, the moderation notes and the record of any adjustment all stay with the institution. 

What travels is a letter or a number per course. A B+ in Organic Chemistry could mean strong lab technique and weak theory, or the reverse. It could come from a supervised practical or from take-home coursework. The transcript doesn’t say, and the reader has no way to ask. 

This is the same compression Part 1 warned about, applied for life. A score that wasn’t evidence on the day it was awarded doesn’t become evidence by being printed on letterhead. 

When the summary stops telling people apart 

A compressed record can still be useful if it separates one learner from another. Grade inflation erodes even that. 

In England, the Office for Students found that the share of students awarded first-class honours rose from 15.7% in 2010-11 to 37.9% in 2020-21. After accounting for entry qualifications and student characteristics, 22.4 percentage points of that figure could not be explained. The rate eased to 32.8% the following year, but the regulator said about half of firsts still could not be explained by entry grades or subject. 

Employers have noticed. In the US, the share of employers screening graduates by GPA fell from 73.3% in 2019 to 38.3% in NACE’s Job Outlook 2024 survey, and stayed below 40% for a third straight year in 2025. Nearly 90% of employers in that 2024 survey said they look for evidence of problem-solving instead. 

That is a quiet verdict on the transcript. Readers haven’t stopped caring about learning. They’ve stopped trusting the summary to show it. 

Better records exist. Most fix a different problem. 

The sector has been trying to enrich the transcript for years, and the efforts are real. 

• The UK’s Higher Education Achievement Report (HEAR) adds university-verified prizes, placements and extra-curricular roles to the academic record, in a standard national format. It is not a statutory requirement, and some universities have stepped back: Exeter issued its last HEARs in winter 2025 and has returned to transcripts. 
• The Comprehensive Learner Record (CLR) 2.0, released by 1EdTech in March 2025 and recommended by AACRAO, packages courses, competencies and skills as cryptographically signed W3C Verifiable Credentials that the learner controls. 
• India’s National Academic Depository, run through DigiLocker alongside APAAR IDs and the Academic Bank of Credits, gives institutions a 24×7 repository to issue verified mark sheets and degrees that students can share with consent. 

Each of these makes the record harder to fake, easier to share or wider in scope. That matters. Paper records are easy to damage and to forge, as India’s Ministry of Education notes, and checking them has long been slow. 

But none of them, on its own, answers the question in this post’s title. A verified B+ is still a B+. A signed credential that says “completed Module 4” is still a statement about attendance and a grade. The record can only say what a learner can do if the assessment underneath it was built to show that. 

Four things a record needs to say what a learner can do 

A transcript doesn’t need to carry every mark ever awarded. It needs to carry enough that a reader, years later, can tell what was demonstrated and how. In practice that comes down to four things. 

1. Outcomes, not just course titles. Each credit should point to the learning outcomes it actually assessed. “Organic Chemistry, B+” says little. “Designs and runs a multi-step synthesis safely: demonstrated” says something an employer can use. 
2. The kind of evidence behind each outcome. A supervised practical, an oral defence, a timed exam and take-home coursework are not equal evidence of the same skill. The record should say which one stood behind the claim. 
3. When it was shown. Skills fade, and a five-year-old result means something different from last term’s. As we’ve argued before, capability evidence has a shelf life, and a record that hides its age overstates what is current. 
4. Verifiable at the source. The reader should be able to confirm the record hasn’t been altered without phoning the registry. This is the part that NAD and CLR 2.0 already solve well. 

The first three depend on assessment that was designed against clear outcomes and recorded per criterion. That is exactly the evidence Parts 1 to 5 asked institutions to keep. A richer transcript is not a new reporting project. It is what you get when that evidence is kept, and then allowed to travel. 

AI changes what a grade can certify 

The second point on that list used to be a refinement. It is now the centre of the problem. 

HEPI’s Student Generative AI Survey 2025, of 1,041 UK undergraduates, found that 92% now use AI in some form, up from 66% a year earlier. 88% had used generative AI for assessments, up from 53%, and 18% had put AI-generated and edited text into assessed work. More students agreed than disagreed (40% to 34%) that AI-generated content would get a good grade in their subject. 

None of that means students are cheating. Much of this use is legitimate, and HEPI itself argues that institutions should treat it as inevitable and often beneficial. It does mean that a grade for unsupervised written work no longer says, by itself, whose capability it measured. If the transcript doesn’t record the conditions, the reader can’t tell either. 

The answer isn’t to surveil every assignment. As we set out in When AI can produce the assignment, it is to design assessment that values process, judgement and dialogue, and then to record which kind of evidence each outcome rests on. 

AI can help here too. It can map criterion marks to programme outcomes across thousands of scripts, and flag outcomes that were only ever shown in unsupervised conditions. What it shouldn’t do is decide what the record claims. A named academic signs off that a learner demonstrated an outcome, as the principle from Who Grades the Grader? requires. 

Part 5 was about answering one person who asks. Part 6 is about everyone who never will: the employer, the admissions tutor or the regulator who reads the record years later and has to take it on trust. The institutions that serve them well are the ones whose records say what was shown, how and when. 

See how Nirvana Intellect supports academic institutions → · Evaluation & capability intelligence → 

Next in the series: Part 7 looks at what happens when a learner moves between institutions, and whether credit earned in one place can be trusted in another.