All guides

The Administrative Assistant Test: What to Give Candidates Before You Hire

A good pre-employment administrative assistant test is short, supervised and built from real tasks. What to put in it, how to read it, and how to run it.

What the Admin Skills Screen is: a supervised, recorded ten-minute exercise of real admin tasks, with single-use codes, that gives your team a score and a plain description, never a verdict. What the screen covers and how it works · All guides

The best pre-employment test for an administrative assistant is a short, supervised work sample: ten minutes of the actual small tasks the job is made of, taken on a live video call, with the result read by a person rather than a pass mark. It should check whether the candidate can follow a multi-step instruction exactly, hold a few details long enough to use them, fix a wrong date, and apply a simple business rule, because those are the things that decide whether an assistant saves you time or costs you time.

This page explains what belongs in an administrative assistant test, what does not, what sample tasks look like, how to read the results, and how to run one with ScreenYourAdmin.

What an administrative assistant actually does all day

Before choosing a test, list the job. Strip away the industry and the software, and most administrative assistant roles come down to a short set of repeated behaviors:

  • Take an instruction, often given once, often with more than one step, and carry it out in the order asked.
  • Notice the exception: "do this as usual, except for the Henderson account."
  • Keep track of dates, names, amounts and sequences between hearing them and using them.
  • Send things out with no wrong dates, wrong names or wrong figures.
  • Apply the simple rules of the business without being reminded each time.

Everything else, from the scheduling software to the tone of a client email, can be taught to someone who does those things reliably. Very little can be taught to someone who does not. A useful test samples those behaviors directly.

What most "admin tests" measure instead

Search for an administrative assistant test and you will mostly find four kinds of product.

Typing tests. Words per minute matter for a transcription role. For an assistant, a fast typist who drops the second step of an instruction is slower in practice than a moderate typist who gets it right the first time.

Software quizzes. Questions about spreadsheet formulas or word-processor menus show whether someone has used a particular tool. They do not show whether that person will put the right date in the cell.

Grammar and vocabulary tests. Useful if the role writes for clients. They say nothing about whether someone can hold a three-item list in mind or follow a reordered instruction.

Long unsupervised assessments. A forty-minute test taken alone at home produces a score, but you cannot know who took it, whether they had help, or whether the answers were looked up. We cover that problem in detail in why unsupervised online tests tell you nothing.

None of these is useless. They just measure something next to the job rather than the job.

Sample tasks from a short admin work sample

The Admin Skills Screen uses tasks of the following shape. The specific words, numbers, days and names are drawn fresh for every attempt, so the examples below are illustrations, not the items a candidate will see.

A reordered instruction. "Do the following in the order 1, 3, 2, 4: (1) type 'invoice', (2) type 'receipt', (3) type 'ledger', (4) type the total number of words you have typed so far." The candidate who reads what is written produces one answer. The candidate who does what they expected produces another.

A reorder with a skip. A four-step instruction where step three says to skip it if a condition is already met, and step four says to insert a word in front of something already typed. This is the office version of "process these as usual, but skip the deposit step if it has already been taken, and put the rush order first."

A short schedule that disappears. Three day-and-task pairs appear for fifteen seconds and then vanish. Which task was paired with Wednesday? Which task came last in the week? This is the Monday call where you run through the week once and expect it to stick.

A sentence with a wrong day. "The delivery arrives Tuesday, not Thursday." The task states which day is wrong and asks the candidate to retype the corrected sentence. It looks trivial, and it is exactly the error that sends a client to the wrong appointment.

A simple rule. "Type your first name. If it has more than five letters, mark it 'underline'; if five or fewer, mark it 'circle'." The rule is easy on purpose. The skill is applying it at all.

Today's day. "If today is a weekday, answer 'open'; if a weekend, answer 'closed'. Then type today's day of the week."

Small figures carried forward. Add two numbers. Sort three numbers and name the middle one. Several items later: "Which number did you identify as the middle value earlier?" Real admin work arrives interleaved, and an assistant has to hold a figure from the morning until the afternoon.

A four-step instruction from memory. The instruction is shown for fifteen seconds, then hidden, and the candidate reports what they produced.

Each item has its own time limit, generous for the task. Speed is not scored. The whole screen takes about ten minutes.

What to look for in the results

The result arrives as a score out of 17, shown as a percentage, a plain band (Strong, Borderline or Below benchmark) and an item-by-item breakdown with a short description of each item. The band is a reading aid. The breakdown is where the information is.

Instruction items missed, everything else fine. The candidate is careful in routine and loses the thread when the instruction departs from the expected pattern. If your work has frequent or expensive exceptions, that matters a great deal.

Memory items missed, instruction items fine. The candidate reads carefully but does not retain what was said. That person often does well with written instructions and a checklist, and less well with "I'll run through it once on the call."

The correction item missed. Look at what they typed. Leaving the wrong day in, or adding the right day without removing the wrong one, is the habit that produces wrong dates in client emails.

The carried-forward figure blank. A blank usually means the figure was never noticed as something to keep.

Two candidates with the same percentage can have very different profiles, and for an administrative role those are not the same hire. Read the breakdown next to the interview, and use anything that stands out as a question in the next conversation. Our guide to testing attention to detail before you hire goes deeper on reading the correction and twist items.

Why supervision makes the score worth reading

A test result is only as good as the conditions it was taken under. On ScreenYourAdmin, a person from your team is on a live video call for the whole attempt. The candidate shares their entire screen, the supervisor reads a single-use code aloud only once the candidate is visibly ready, and the candidate's camera, microphone and screen are recorded from start to finish, with their written consent. Every attempt draws fresh values, so there is no answer key for a previous candidate to pass on.

That combination is what turns a quiz into something you can rely on. You know who took it, under what conditions, and you can watch the recording if anything in the result surprises you.

How to run an administrative assistant test with ScreenYourAdmin

  1. Create a code. Sign in, type the candidate's name, and create a single-use code. If the candidate has asked for extra time because of a disability, set 1.5x or 2x time on the code.
  2. Send the setup note. The candidate needs a laptop or desktop, Chrome, and the video-call link in a second tab. They need no account and install nothing.
  3. Run the call. On the video call, check the candidate is alone with their whole screen shared, read the code aloud, and stay on the call while they work. The candidate signs the consent, enters the code and starts.
  4. Read the result. The score, band, item breakdown and both recordings appear on your dashboard as soon as the candidate finishes. The candidate is told the exercise is complete; they never see a score.
  5. Decide. Your team makes the decision. The software never accepts or rejects anyone.

The minute-by-minute guide to running the screening call covers the call itself in detail.

What the test will not tell you

Be clear with yourself about the limits. A ten-minute admin work sample does not measure typing speed, knowledge of a particular software package, phone manner, writing style or whether someone will enjoy the role. If those matter for your position, check them separately: a short writing task, a call with a client-facing member of the team, a paid trial day. The screen covers the core habits that every one of those other skills depends on, and it is designed to be one input among several, never the whole decision.

If the role is mostly bookkeeping rather than general admin, there is a separate Bookkeeping Skills Screen covering bank-feed coding, journal entries, a bank reconciliation and month-end adjustments, taken under the same supervised conditions.

Using it fairly

Give every candidate for the role the same test, under the same conditions, at the same point in the process. Tell them in advance what it is: a ten-minute exercise of everyday admin tasks, on a video call, recorded, one attempt, nothing to study. Offer extended time to anyone who asks for an accommodation, and do not penalize them for asking. Never let a score be the only reason for a decision. Those habits make the result more useful and the process fairer, and our overview of pre-employment testing legal basics explains why consistency and job-relatedness matter.

Try it before you use it on candidates

Many employers take the screen themselves, or run it on a trusted member of the current team, before sending it to a candidate. It takes ten minutes and gives you a feel for what a clean breakdown looks like in your own context. The values you see will never appear for a candidate.

Credits are sold in packs of 5, 20 and 50, with no subscription; one credit is one supervised attempt, on either the admin or the bookkeeping screen. See the credit packs on the pricing page, or read how the screen works on the home page.

Frequently asked questions

How long should an administrative assistant test be? Short. The Admin Skills Screen takes about ten minutes. The skills being checked are basic, and beyond ten or fifteen minutes a test starts measuring stamina rather than the job.

Can a candidate take the test on their phone? No. The candidate needs a laptop or desktop computer and the Chrome browser, with the video call open in a second tab. Phones are not supported.

Does the candidate see their score? No. Candidates are told the exercise is complete and that you will be in touch. The score, band, breakdown and recordings go to your dashboard only.

Who on my team can see the results? The account owner sees every score, recording and signed consent. A proctor, invited by the owner to run calls, can create codes and see that an attempt completed, but not the score.

Can candidates share the questions with each other? They can share the format, and we tell candidates what to expect. Every attempt draws fresh values and every code works once, so there is no answer key to pass around.

Is this a typing test? No. Each item has a generous time limit and speed is not scored. If typing speed genuinely matters for your role, test it separately.

The Admin Skills Screen

A supervised, recorded, ten-minute screening exercise for remote admin hires. Single-use codes, fresh content every attempt, one input to your decision.

How the ten-minute screen works →
Credit-pack pricing, from five candidate credits