Graduate employability: what employers want, and what holds them back

The largest panel in GEURS history

Between August and September 2026, we surveyed employers in 40 countries. By the time we had removed the search-engine bots, the people who opened the survey and typed nothing, those the survey itself screened out, and the ones who raced through so fast they cannot have read the questions, we had 15,665 verified employers: the largest panel in GEURS history.

Data: the four checks behind the 15,665 verified employers

Check	Treatment
Search-engine bots	Removed
Empty visits	Removed
Out of scope	Screened out: not in the target population
Speeders	Answered too fast to have read the questions

GEURS 2027 employer survey, fielded 17 Aug – 7 Sep 2026, 40 countries.

Who answered

They are a serious group. Nearly two thirds have more than ten years of experience. Seven in ten have personally worked with at least ten graduates in the past two years. On average, respondents rate their influence over hiring criteria at 6.8 out of 10.

Data: profile of the respondents

Indicator	Value
More than ten years of professional experience	64%
Worked with at least ten graduates in the past two years	69%
Average influence over the criteria candidates are hired on	6.8/10

The common core

Take one question as an example: which practical skills matter most when you hire a graduate? Across eight world regions, the general tendency is clear. Technical ability comes first overall, digital fluency second, and cross-border, cross-cultural experience last, a last place it holds in every region.

This common core shows up across the survey. We looked at every question by country, industry, company size, department and AI maturity. The general tendencies hold everywhere, and each segment adds its own emphasis: that is where the detailed results by country and sector come in.

Data: practical skills that matter most when hiring a graduate, all regions (share of selections)

Skill	Share of selections
Technical ability	25.2%
Digital fluency	22.9%
Discipline expertise	20.3%
Industry experience	19.3%
International experience	12.3% (last in all eight regions)

Share of selections, so groups that tick more boxes are not over-counted.

What employers actually want

Here are the tendencies worth knowing first. Start with AI. Ask around and you will hear that employers are desperate for prompt engineers and Python developers. Our data points to a different order of priorities. Asked which AI skills are in highest demand in their organisation, employers named implementation and integration first (30.7%), then strategy and use cases (28.8%), then training their own people (28.6%).

Prompt engineering came sixteenth of seventeen. Python for AI came last.

Data: AI skills in highest demand, three most and three least cited of seventeen

AI skill	Rank	Employers
AI implementation & integration	1st	30.7%
AI strategy & use cases	2nd	28.8%
Internal AI training & upskilling	3rd	28.6%
Human-in-the-loop design	15th	15.9%
Prompt engineering	16th	15.5%
Python for AI	17th	13.4%

Employers could pick as many as applied.

What holds AI back

Read the top of that list again and a pattern emerges: every one of those skills is organisational. Their leading priorities concern putting AI to work across the organisation. They are short of people who can get a model into a business that was not built for it, and keep it compliant once it is there. Their stated barriers say the same thing. Cyber security (32.3%) and worries about reliability and hallucinations (28.5%) rank far above budget (20.2%).

Data: barriers to AI in the organisation

Barrier	Rank	Employers
Cyber security & AI-specific risks	1st	32.3%
Reliability of outputs (hallucinations)	2nd	28.5%
Lack of AI skills internally	3rd	25.8%
Budget constraints	9th of 13	20.2%

Employers could pick several barriers.

The contradiction

Then there is the contradiction at the heart of graduate hiring. We asked employers to rank four things by how decisive they are. Formal qualifications came last: 34.9% put credentials at the bottom. Then, a few questions later, we asked which signals actually matter when they look at a candidate. Internships and work exposure came first, at 65%. Degrees and certifications came second, at 59.5%.

Both answers are true, and together they describe how hiring really works. Qualifications remain an important signal. Practical experience helps candidates stand out. The internship decides what happens next. A qualification is necessary and almost never sufficient, which is exactly why employers rank it as the least decisive thing they look at while still requiring it.

Data: ranked least decisive when hiring, and the signals that matter most

Answer	Question	Employers
Credentials	Ranked least decisive	34.9%
Meta skills	Ranked least decisive	25.4%
Transformation	Ranked least decisive	22.2%
Execution	Ranked least decisive	17.5%
Internships & work exposure	Top signal	65.0%
Degrees & certifications	Top signal	59.5%
Networking	Top signal	38.0%

The starting point

Nine in ten employers told us they have some kind of relationship with a university. Only 18.2% described it as deep and structured. Most of the rest is occasional: a guest lecture, a careers fair, an internship pipeline that nobody owns.

Data: how employers describe their engagement with universities

Engagement	Employers
Deep and structured	18%
Several ongoing activities	38%
A few occasional activities	34%
No active cooperation	10%

AI and cooperation

And one of the clearest dividing lines between deep and occasional partnerships is AI.

Among employers who describe themselves as AI-native, 45.5% have deep, structured cooperation with universities. Among employers not using AI at all, it is 6%. Seven and a half times the rate.

Data: deep, structured cooperation with universities, by AI maturity

AI maturity	Employers
No AI use yet	6.0%
Exploring & piloting	11.7%
Using AI in some functions	11.2%
AI embedded across business	20.8%
AI-native	45.5%

The AI partner

It is one of the strongest relationships in the survey, and it runs in a direction universities should find encouraging: when employers decide to build AI capability seriously, universities are the partner they name first: 52.1% say they are well prepared to build AI skills with a university, narrowly ahead of cloud and AI vendors at 51.9%.

Data: well prepared to build AI skills with each type of partner

Partner	Employers
Universities & business schools	52.1%
Cloud & AI vendors	51.9%
Specialised AI training providers	46.5%
Online platforms & MOOCs	44.5%
In-house academy	42.6%
Professional bodies & certifications	40.9%

Beyond borders

AI is not the only dividing line. Employers who recruit at global level are four times as likely as those who recruit only at home to have a deep, structured partnership with a university (34.3% against 8.3%), and almost none of them work without universities at all (3%, against 22%). The gap holds for companies of the same size and across levels of AI maturity.

These employers are also the most satisfied with their cooperation (59% very satisfied, against 31%) and the most ready to build AI skills with a university (65%, against 39%). The more widely an employer recruits, the more it invests in universities.

Data: by how far employers recruit graduates

Recruitment reach	Deep, structured partnership	Very satisfied with cooperation	Ready to build AI skills with a university
Recruit only at home	8%	31%	39%
Also recruit from abroad	16%	36%	53%
Recruit at global level	34%	59%	65%

What gets in the way

But the same employers are clear about what gets in the way. Budget and time come first (26.1%), followed closely by two gaps on the graduate side: a lack of practical experience and a lack of AI and digital capability (25.1% each), ahead of administrative complexity (22.1%).

Data: what limits the quality of cooperation with universities, five most cited factors

Factor	Employers
Budget or time	26.1%
Graduates lack AI & digital capability	25.1%
Graduates lack practical experience	25.1%
Universities slow to adapt to technology	23.1%
Administrative complexity	22.1%

Employers could pick several factors.

The money question

Notice the contrast with AI. When employers talk about adopting AI, money is a secondary concern: budget ranks only ninth of thirteen barriers. When they talk about working with universities, budget and time come first. For AI, the hurdles are trust and skills; for partnerships, they are budget and time. That is an opening for universities that make cooperation simpler and quicker to set up.

Data: where budget ranks among the barriers

Context	Rank of budget	Employers
Adopting AI	9th of 13 barriers	20.2%
Cooperating with universities	1st of 20 barriers	26.1%

Two separate questions; employers could pick several barriers in each.