Graduate screening: how to assess skills beyond the CV

One graduate has an impressive internship. Another spent weekends checking deliveries in a shop. You need a junior analyst who notices when a number does not make sense.

Ask both how they checked a number that looked wrong. A delivery note and a spreadsheet can each reveal careful reasoning; an employer’s name cannot tell you what the applicant actually did.

Assess the skills the graduate needs on day one, and be clear about what you will teach. A short analyst exercise can compare calculations, judgement and responses to new information without requiring an office internship.

How do you assess graduates with little work experience?

Start with what the person must be able to do before training. Then choose a question or task that gives applicants a chance to demonstrate it, even without an internship. Ask the manager:

Which skills must the new hire already have? Who will help them learn the rest during their first month?

For the junior analyst role used here, accurate basic calculations and careful reading are needed on arrival. The company will teach its reporting software and definitions. Name who will provide that training and review the new hire's work.

Confirm that support before advertising. If nobody can train a beginner, the team may need an experienced hire. That is a limit on the support available, not proof that a graduate without office experience lacks ability.

The US Office of Personnel Management’s work-sample guidance distinguishes skills applicants need on entry from work the employer will teach after hiring. Here, assess arithmetic and interpretation, not familiarity with your reporting software. Removing an internship requirement does not remove the need to calculate accurately.

Ask for examples from work, study or volunteering

An early-career applicant may have a relevant example but assume you only want office work. Explicitly welcome paid work, study and volunteering, then ask:

Tell me about a time a number looked wrong. How did you check it?

The shop worker might have checked a delivery against the order. The intern might have compared an invoice with a spreadsheet. Ask each what looked wrong, how they checked it and what they found. Compare the checking method, not the employer's name.

The figure need not have been wrong. Checking carefully and confirming it was correct may show exactly the habit the job requires. You are assessing the method, not looking for the most dramatic rescue story.

Use the task when the application and conversation leave this skill unclear. Agree how you will choose who takes it. Apply that rule consistently: an impressive employer name should not exempt one applicant from demonstrating the skill while another must complete an extra test.

Graduate analyst exercise: which support team needs help?

Two support teams have customers waiting. Which team needs help first? Give the applicant the facts in three stages and ask them to explain their recommendation as they learn more. They do not need to guess information you have not supplied.

This exercise is not a validated selection test. Try it with your manager before using it in hiring, and agree the time allowed, permitted tools and adjustment arrangements. Allow a calculator when you want to assess interpretation rather than mental arithmetic. Give everyone a chance to check their work before calling an answer final, and keep the answer notes in the interviewer's guide.

Stage 1: show the day's activity

Give the applicant this brief and table:

At the end of a working day, you have these figures for two support teams. What would you check before recommending where help is needed?

A useful answer might ask how many requests were already waiting, whether any are urgent or how many staff are available. If the applicant says only “we need more data,” ask: “What would you ask for first, and why?”

Supply the backlog and urgency facts when asked. Otherwise, introduce them during the exercise. Note who asked independently, but make sure everyone has the same facts before giving their final recommendation.

Leave staffing and other unspecified details unknown. An improvised answer during one interview would change the task for that applicant.

Can you compare the teams using percentages?

Dividing requests resolved by new requests gives about 67% for Team A and 83% for Team B. That doesn't prove Team B is more productive. The resolved requests may include older work, and the table says nothing about difficulty or staffing.

If an applicant calculates these percentages, ask what they mean before deciding the answer is wrong. They may understand the limits and simply be showing one calculation.

Stage 2: add the starting backlog

Tell the applicant that Team A started with 20 requests waiting, and Team B started with 1. New requests arrived and resolved requests left the queue. None were transferred or reopened.

Ask how many requests remained at the end of the day. Interviewer answer guide:

Team A has 24 requests left and Team B has 2. Those totals show where more requests are waiting. They do not show which team is less productive: staffing, request difficulty and available working time are still unknown.

The applicant can make a sensible first recommendation without pretending to know everything: “Team A has more requests waiting. I'd check how urgent they are before deciding which team needs help first.”

Stage 3: reveal the urgent customer problem

If it has not already been requested, add this fact:

“At the end of that day, one of Team B's two open requests concerns a customer whose service isn't working. Team A's remaining requests are all lower-impact issues.”

Team B now has the urgent problem, despite its smaller queue. Listen for an applicant who recognises that priority and asks who can help. Immediately moving staff from Team A would assume that somebody there has the right skills and capacity.

Once everyone has received the same facts, ask for a short recommendation. A satisfactory answer might be:

Team A has 24 requests left and Team B has two. I'd first find someone who can help the customer whose service is down. Then I'd check how long Team A's requests have been waiting and what work they need. I wouldn't move staff between teams until I knew who was available and had the right skills.

Accept a different plan if it addresses the urgent customer and the larger backlog. The applicant should explain what they would check before moving staff, rather than assume that anyone available can resolve the problem.

How should you score a graduate assessment?

Record the calculation and recommendation separately. Correct arithmetic can support a poor conclusion; a corrected arithmetic slip can accompany sound reasoning. Note whether the applicant found the mistake independently, during the agreed checking time or after a hint. The manager needs to see both the work and the help it required.

Distinguish task facts from help with the reasoning. Supplying the urgent-customer detail lets everyone complete the task. Explaining why the percentages do not establish productivity teaches part of the answer. Agree those prompts beforehand and record when you use them.

Ask how the new information affects their plan. They may refine a good first answer rather than reverse it. Assess whether the plan makes sense, not whether they change their mind on cue.

Your assessment can stay focused:

Compare the work, not a general impression of potential

Use these practice responses to agree the standard before interviewing:

Have two reviewers score these responses independently. Discuss how a corrected arithmetic error and an unsupported productivity claim affect this role's standard. Agree the reasons before assessing applicants.

Record what the prompt supplied and what the applicant worked out afterwards. “Explained why backlog size does not establish productivity after the interviewer asked about staffing” is more informative than “fast learner”. One improved answer does not establish a general rate of learning.

Optional follow-up: can they use the idea on new numbers?

Use this only when applying feedback is relevant to the decision. Team C starts with 7 requests, receives 4 and resolves 3. Team D starts with 4, receives 6 and resolves 5. Nothing is transferred or reopened. Ask for the remaining work and a recommendation.

The totals are 8 for Team C and 5 for Team D. Leave urgency, staffing and difficulty unspecified. Does the applicant now distinguish the size of the backlog from a claim about productivity? Do they ask for a fact that would help decide where assistance belongs?

Use the same follow-up and permitted help for comparable applicants. Record it separately from the first attempt. It shows how they applied this lesson on another example, not a general prediction about their future learning.

Reserve time to review the task before assigning it

Decide who will read the work before assigning it. Give reviewers the same facts and rating guide, and reserve time to resolve disagreements. Check existing evidence first: do not assign another exercise merely to make the shortlist smaller through extra candidate effort.

Record independent work, prompts and training needs

Replace “high potential” with a description of the work. The manager needs to know what the applicant handled independently, what needed a prompt and what remains to learn.

For this vacancy, the company will teach its reporting software. Record whether the applicant could calculate the waiting work, spot the urgent request and explain what else they needed to know. They do not need specialist vocabulary to do that.

A note about an answer that improved could read:

The applicant calculated both backlogs correctly. They first used the percentages to compare the teams' productivity. After one prompt, they explained why that comparison was incomplete. They prioritised the customer without service once the urgency information was added.

Match that observation to the entry standard and the training the team can actually provide. The manager can then judge the initial mistake and the response to help instead of inferring both from “high potential”.

Mark an essential skill not assessed when the interview never reached it, then arrange the missing assessment. Keep that gap separate from an answer that fell below the standard.

Return to the two applications. The internship and shop job gave you places to look for evidence; neither supplied the hiring decision. Now you can point to how each applicant calculated, checked and revised a recommendation. Assess the work against the agreed entry standard, then name the training and support the successful hire will receive.