Job adverts describe outcomes. Nobody describes the hours#
Every graduate job advert is written in the same register. Drive insight. Deliver impact. Partner with stakeholders. Own the roadmap.
None of that tells you what Tuesday is like, which is the thing you are actually choosing. People do not leave jobs because the outcomes were wrong. They leave because the days were.
So here are four common graduate destinations, described by what the hours contain rather than what the work is for. Every one of them has a substantial unglamorous majority that nobody mentions at a careers fair, and in each case that majority is the job.
The pay context, because it frames all four#
UK graduate salary headlines are quoted selectively, and the gap between the figures is large.
Median starting salary at the UK's top 100 graduate employers is expected to hold at £35,000 in 2026, according to High Fliers research, ending five years of consecutive increases. The Institute of Student Employers reports an average of £36,335 across its member employers.
Those figures describe a specific slice of the market. The median for all UK first-degree graduates in full-time work fifteen months after leaving is £28,500, based on HESA Graduate Outcomes data, and broader market estimates put the 2026 average around £30,500. Most graduates join smaller organisations paying between roughly £22,000 and £28,000.
Data analyst: mostly cleaning, occasionally analysing#
The advert says you will uncover insight from data. The week says you will spend a surprising amount of it working out why two systems disagree about the same number.
A realistic day involves writing SQL against a warehouse, fixing something in a dashboard, and answering a question from someone in another team who wants a figure by Thursday. A significant share of the hours go on data cleaning and reconciliation, which is the work of getting messy real-world records into a state where a calculation means anything.
The recurring conversation is not statistical. It is definitional. Somebody asks how many active users there were last month and it turns out three teams define active differently. Resolving that is the job, and it is more political than technical.
The part nobody warns you about: you will often be asked to produce a number that supports a decision someone has already made. Handling that honestly, without either capitulating or being difficult, is a career skill in itself.
Who does well: people who enjoy precision, tolerate ambiguity in requirements, and can explain a finding to someone who does not want to hear it. Who struggles: people who expected modelling and find themselves in spreadsheets.
Entry-level pay sits at roughly £25,000, with graduate schemes at larger employers higher. Common next steps are senior analyst, analytics engineer, data science, or moving into the business function you have been supporting.
Product manager: mostly writing and deciding what not to do#
Product management has the most misleading job title in technology. You do not manage anybody, and you are not the boss of the product.
The day is meetings and writing, in roughly that proportion. You talk to engineers about what is feasible, to designers about what it should feel like, to sales or support about what customers complain about, and to leadership about why the thing is not shipped yet. Then you write it all down, because a decision that only exists in a conversation did not happen.
The real work is prioritisation, which mostly means saying no. Every stakeholder has a request that is genuinely reasonable, and you have one team. Most of the difficulty in the role is disappointing people while keeping them onside.
Very few organisations hire graduates directly into product management, because the role runs on credibility you do not have yet. The usual routes in are through an associate product manager scheme at a larger technology company, or sideways from support, analytics, engineering or marketing after a year or two.
Who does well: people comfortable with responsibility without authority, who write clearly and do not need to be the cleverest person in the room. Who struggles: people who want to build the thing themselves, and people who need decisions to be provably correct.
Management consultant: slides, models, and other people's problems#
Consulting is the most structured of the four and the most demanding on hours.
The graduate reality is research, analysis and slide production. You will build spreadsheet models, interview client staff, synthesise what you find, and turn it into a deck that a partner presents. Formatting matters more than you would believe, because the deck is the product.
Projects run in blocks of weeks or months, you are staffed onto them rather than choosing them, and the subject changes completely each time. That variety is the genuine attraction: you will see more industries in three years than most people see in twenty.
The costs are equally real. Hours are long and unpredictable, travel is common though less than it once was, and you spend a lot of energy on internal positioning as well as client work. Attrition is high and largely by design, since the model assumes most juniors leave.
What you get out of it is a reputation that opens doors, a network, and a way of structuring problems that transfers well. The standard exits are into strategy or operations roles at client companies, into finance, or into startups.
Who does well: people who work fast, take feedback without collapsing, and are comfortable being the least experienced person in a client room. Who struggles: people who want depth in one subject, and people for whom unpredictable hours are genuinely unsustainable.
Software engineer: writing code is the minority of the job#
The surprise for most computer science graduates is how little of the week is spent typing new code.
The rest is reading code somebody else wrote, understanding a system nobody documented, reviewing colleagues' changes, debugging something that only fails in production, waiting for builds, and sitting in planning meetings. Debugging in particular is a much larger share of the job than any degree implies.
Work arrives as tickets, gets estimated, gets built, gets reviewed, gets deployed, and occasionally breaks at an inconvenient hour if your team has an on-call rota. Code review is where most of the learning happens in the first year, and it is also where people are most likely to feel exposed, because your work is examined publicly by default.
The best first-year engineers are not the fastest coders. They are the ones who ask a clear question after twenty minutes of being stuck rather than three days.
Who does well: people who enjoy diagnosis, tolerate the tedium of existing systems, and can accept critical review of their work as normal. Who struggles: people who liked programming for the greenfield building and find themselves maintaining something ten years old.
Pay varies more than in any of the other three, from ordinary graduate scheme levels at most UK employers to substantially higher at large technology companies and some financial firms.
The question to ask at an interview#
For any of these, the question that produces the most honest answer is not what a typical day looks like. It is: what did you spend most of last week doing, specifically?
People will answer that truthfully, because it is a question about memory rather than about the company. And if the answer sounds tolerable, that tells you far more than any job description.
A second one worth asking: what does someone in this role who is doing badly get wrong? The answer reveals what the role actually rewards.
How to test-drive any of these before committing#
All four are easier to sample than students assume, and a weekend of doing beats a month of reading.
For analytics, download a public dataset and answer one question with it, properly, including the part where the data turns out to be messier than advertised. For engineering, fix an open issue in a small open-source project. For product, write a two-page specification for a change to an app you use daily, including what you would deliberately not build. For consulting, take a company you know and write one page structuring a problem it has.
Each takes an afternoon and tells you more than any number of careers-fair conversations, because what you are testing is whether the activity itself is tolerable rather than whether the outcome sounds impressive.
Frequently asked questions#
Which of these pays best at entry level? Software engineering and consulting tend to top the range, with large technology companies and some financial firms paying well above the top-100 median of £35,000. Data analyst roles typically start lower, around £25,000, and product management is rarely an entry-level role at all.
Do I need a technical degree for data analytics? No. Economics, geography, psychology, biology and many other quantitative-adjacent degrees are common. SQL and a spreadsheet tool matter more than your subject, and both are learnable in a few months.
How do people become product managers? Rarely straight from a degree. Most arrive through an associate product manager scheme at a large technology company, or move across from support, analytics, engineering, design or marketing after one to three years.
Is consulting worth it if I do not want to stay? That is the standard case rather than the exception. The model assumes high attrition, and the exit options are a substantial part of what you are buying. Go in with a rough view of how long you intend to stay.
Will AI tooling remove these entry-level jobs? It is changing what the junior work consists of rather than removing the roles wholesale, and the picture varies by field. The parts most exposed are the routine, well-specified tasks. The parts least exposed involve deciding what is worth doing and whether an output is actually correct, which is where a junior should be building capability anyway.
How much do hours really vary between them? Consulting is the most demanding and least predictable. Engineering and analytics are usually closer to standard hours, with exceptions around releases and incidents. Product management sits in between and tends to be meeting-heavy rather than late.
Can I switch between these later? Yes, and people do constantly. Analyst to product manager, engineer to product manager, and consultant to almost anything are all well-trodden routes. Your first job is a starting point, not a classification.
