Taking an assessment rather than buying one? This page is written for employers. Here is the page for candidates.

Administrative and business support · Mid level

How to assess a Office Manager

Office manager is a title that means five different jobs — in one company it is senior administration, in another it is de facto HR and finance for a fifty-person business — so a screen that asks whether the candidate has held the title before selects for having done a possibly unrelated version of it. The role's real exposure is that it holds a discretionary budget, signs or renews supplier contracts, and hears things before HR does, all with very little supervision, and neither the money judgment nor the discretion judgment is examined anywhere in a normal process. Candidates are asked how they would handle a difficult supplier; they are never shown a renewal notice, a price increase and a usage report and asked what they would actually do about it, which is a question with a right answer and a gradeable one.

A scope note belongs at the top of this one, because the corpus is deliberate about what it will not claim. Part of an office manager's job is physically present: the space, the front of house, the fire drill, the person who arrives with a delivery. None of that can be observed remotely, and nothing in this corpus pretends otherwise — the taxonomy excludes reception, front-desk and facilities work by name. BLS draws the same line in its own data, reporting administrative services managers (279,000) separately from facilities managers (165,900). What is left is the half of the role that is spend, suppliers, policy and people administration, and it is both the larger half in most organisations and the half where the expensive mistakes happen.

The defining structural feature is discretion with very little supervision. An office manager in a growing company typically holds a real budget, is the named account contact for a dozen suppliers, administers benefits and onboarding, answers questions about policy that nobody has written down yet, and reports to a finance director or a founder who has neither the time nor the inclination to review any of it. Nobody checks the software renewal. Nobody re-reads the cleaning contract. The role is trusted by default because scrutinising it would cost more attention than the line items appear to be worth, which is precisely how a fifty-person company ends up paying for ninety seats of a tool.

Supplier judgment is therefore the first thing that separates a strong hire. The characteristic weak behaviour is not overspending; it is renewing. A notice arrives sixty days before the anniversary, it is filed, and the auto-renewal clause does the rest — and the following year the same thing happens with a seven percent uplift attached. The strong behaviour is a small, unfashionable discipline: knowing the notice dates, pulling the usage before the conversation, and treating a renewal as a decision that has to be re-made rather than a payment that has to be approved. It is entirely gradeable from an artefact, because the recommendation either references what the organisation actually consumed or it does not.

The second differentiator is applying a rule to a case it does not quite fit. Office managers are the first port of call for questions the policy does not answer — an expense that is borderline, a request to work a pattern nobody has worked before, an equipment claim from someone who has already had two. The failure mode is bimodal and both halves are common: quoting the policy back at someone whose situation the policy did not anticipate, or waiving it informally and creating a precedent that the next person will cite. The competent middle is to decide the case on its facts, say which part of the decision is discretionary, and route the ones that should not be theirs to decide.

That last clause matters enough to state plainly, because it is the highest-risk thing this role does. An office manager frequently becomes the person a colleague tells something to first — about their health, their home, or a grievance about a manager. The correct behaviour is entirely about routing and confidence: recognising that this is not theirs to hold, saying so kindly, and getting it to the right person without broadcasting it. Any assessment of this must grade the routing and the discretion only, and must not put the candidate in a position where they are judged on their attitude to the underlying circumstance. Assess whether they knew it was not their decision. Never assess what they would have decided.

The last thing a hiring manager is trying to predict is whether the person will fix processes or absorb them. Every office manager is offered the choice weekly: handle the request personally, which is faster today, or change the thing that generates it, which is slower today and compounds. Strong hires are visible in what they choose to write down.

What the job actually needs

• supplier and contract judgment
• budget variance ownership
• applying policy to an individual case
• writing to a whole organisation
• prioritising between people who all outrank the role

How people fail in this seat

• renews a contract on autopilot after missing the notice period
• applies a policy to a sensitive personal situation by quoting it back
• spreads a discretionary budget evenly rather than where it matters
• absorbs every request personally instead of fixing the process behind it
• routes a confidential disclosure to the wrong person

What most employers do instead

CV screen for prior tenure in the same title, a conversational interview, and references.

The assessment

About 38 minutes end to end.

The systems it runs in

Four surfaces, all of them on the systems side of this job and none of them on the site. A budget-against-actuals export from the finance system, which is where the training provider's overspend is visible and where the renewal uplifts have to land. The contract records themselves, as the renewal packs and covering emails actually arrive — which is the point of the notice date hiding in the third paragraph of an email rather than in a field called notice date, because in most organisations of this size there is no field. A seat or licence usage export from the software supplier's own admin console, which is the only place the gap between ninety contracted seats and the number of people actually signing in can be seen. And a spend or expense platform — Ramp and Expensify are the named reference builds — which is where the over-limit claim from a senior colleague arrives with its receipt and its stated business reason, and where a decision on it is recorded as an approval or a rejection rather than as an opinion. The spreadsheet work is the renewal analysis and it is a specific job, not a formula exercise. The seat export lists sign-ins by user account; the contract lists ninety seats and a per-seat price; the two do not share an identifier, because the export keys on email address and the contract keys on nothing at all. Turning that into a defensible recommendation means counting distinct active accounts over the quarter rather than rows — the export has one row per session, so a naive count returns a number several times too large and in the direction that argues for renewing at ninety. Then it means stating what that count implies about the right seat number and pricing the difference against the seven percent uplift. Fast and correct is a candidate who notices the export is per-session before summarising it; fast and wrong is a row count, which looks like analysis, supports the easy recommendation, and is the only version that never gets challenged. What is deliberately absent is as important. There is no facilities system, no visitor management, no access control and no maintenance queue in this environment, because the physically present half of this job is not assessed here and a tooling list implying otherwise would be the easiest way to let a buyer read the score as covering it. There is also no HR or people record. The colleague who discloses something about their own circumstances is routed and nothing more; the fixture gives the candidate no personnel file to open, no field to record it in, and no system in which the substance could be written down, which is the design making the constraint structural rather than relying on a reviewer to honour it.

• Microsoft Excel
• Google Sheets
• Ramp
• Expensify
• Sage Intacct
• Oracle NetSuite
• Microsoft 365
• Google Workspace

Any finance system that produces a budget-against-actuals export, any corporate card or expense platform with an approval step, and any software vendor's admin console or usage report. Where the buyer runs a contract or procurement system with renewal dates as structured fields, the fixture is rebuilt against it — and the file notes that this makes the first task easier rather than harder, because the notice date stops being buried, so a buyer who has one should expect the scores on the first criterion to rise and should recalibrate rather than compare across builds. No criterion mentions a platform, and none is satisfied by knowing where a menu item lives.

What the candidate actually does

The mark scheme

Each criterion is scored 1 to 5 against written anchors, and every score is reported with the excerpt that earned it. A criterion marked floored is reported as a finding rather than averaged into the total. The first is open; open any other to read its anchors in full.

How it is scored

Weighted mean of the five criteria, each scored 1 to 5 against the anchors and reported with the contract line, card answer or sentence that earned it. Two figures are reported beside the score and annotated explicitly as not positive signals: the total spend the candidate recommended approving, and the number of the three suppliers they proposed to challenge. Neither is a result. A candidate who challenges all three has failed the proportion the fifth anchor of the first criterion describes, and a candidate who recommends the lowest spend may have done so by cancelling something the organisation needs. The report also states whether the notice deadline appeared anywhere in the candidate's output, as a plain yes or no, because it is the single most consequential detail in the fixture and it is either seen or it is not.

Integrity

• one unbroken monitored sitting, with the supplier analysis committed before the judgment cards open
• both written documents are graded against the candidate's own analysis, so a prepared renewal template cannot be made to fit
• fixture variants rotated between sittings so the identity of the over-provisioned contract, the position of the notice date and the content of the borderline expense differ
• one live follow-up question at the mid band, which invalidates an assumption rather than adding work: the candidate is told the supplier has replied that the seat count cannot be reduced mid-term and the uplift is non-negotiable, and is asked what they now recommend and what they would tell the finance director
• session log records file-open order across the three renewal packs, time distribution and paste-versus-typed provenance; no automated integrity verdict is produced from it

The log describes what happened. It does not produce a cheating verdict — the follow-up conversation is the control, because a statistical accusation is not something we would ask a reviewer to defend.

What you receive

• the three supplier recommendations with the usage or spend evidence cited for each, and the stated budget position
• the recommendation to the finance director and the supplier message as submitted
• the six card answers with their one-line reasons
• per-criterion score with the contract line, card answer or excerpt that earned it
• recommended spend and number of suppliers challenged, annotated as not positive signals, and a yes or no on whether the notice deadline was identified

Who decides

Recommended. The reviewer has one decision the rubric deliberately leaves open: where this organisation draws the line between what an office manager decides and what they route. That line sits in a different place in a fifty-person company with no HR function and in a five-hundred-person one with a full people team, and the same disposition on the working-pattern request or the expense claim is correct in one and overreach in the other. The reviewer states the buyer's own boundary before scoring and overrides the third criterion in writing where it differs — never the fourth, which is not a boundary question. The ranking entitles a buyer to conclude that this candidate treats a renewal as a decision to be re-made, can apply a rule to a case it does not fit without either quoting it back or quietly waiving it, and knows what is not theirs to hold. It does not establish anything about the half of this job that happens in a building.

What this does not measure

This design covers only the part of the role a remote session can honestly observe: spend, suppliers, policy applied to individual cases, and people administration. The physically present half — the space, the front of house, the fire drill, the delivery at the door — is not assessed and no score here should be read as evidence about it. That is a scope limit stated in the role page and it is repeated here because a buyer looking at a single number will otherwise assume it covers the whole job; the taxonomy excludes reception, front-desk and facilities work by name, and BLS itself reports administrative services managers separately from facilities managers, so the line is drawn in the source data rather than only in our copy. A deployer who needs the on-site half assessed should assess it on site. The confidential disclosure card is scored on routing and discretion only. The fixture does not state what the colleague's circumstances are, the candidate cannot ask, and no anchor rewards or penalises any view of them: what is graded is whether the candidate recognised that it was not theirs to hold, told the person what would be passed on before passing it, and kept the substance out of the artefacts. A reviewer who finds themselves reasoning about the underlying circumstance has left the rubric, and a design that graded the candidate's attitude to it would be screening on exactly the kind of thing this corpus refuses to screen on. The supplier analysis is presented as plain exports and contract extracts, any method of arriving at a figure is accepted, and no spreadsheet formula use is scored. What the design cannot reach, beyond the on-site half, is the pattern over a year: the failure that costs the most in this role is not one missed notice date but the habit of filing them, and one folder samples the habit without observing it repeat.

A scope note belongs at the top of this design, because it governs what the assessment is entitled to claim. Part of an office manager's job happens in a building and cannot be observed remotely, and nothing here pretends otherwise. What is assessed is the other half — the spend, the suppliers, the policy applied to individual cases, the things people bring to this person before they bring them to anybody else — which is the larger half in most organisations and the half where the expensive mistakes happen. A buyer who needs the on-site half assessed should assess it on site, and a score from this session says nothing about it.

The defining structural feature of the role is discretion with very little supervision, so the fixture reproduces the absence of supervision rather than simulating a manager. The finance director in this brief has said, in the same words for every candidate, that they do not want to be involved in the detail and will approve what is recommended. That sentence is the whole design. Nobody checks the software renewal. Nobody re-reads the contract. The role is trusted by default because scrutinising it would cost more attention than the line items appear to be worth, which is precisely how a fifty-person company ends up paying for ninety seats of a tool.

The renewal pack is built to punish both failure modes. Approving all three is one action and generates no friction with anyone; the characteristic weak behaviour in this role is not overspending but renewing, and the fixture makes renewing frictionless. But challenging all three is also wrong, and the second supplier exists solely to catch it: small, cheap, universally used, renewing flat, and correctly disposed of in one line. A candidate who builds a case against it has misunderstood what their attention is for, which is the same proportion error the third supplier tests from the other direction — its spend has run over budget, its invoices do not match the rate card, and the service is genuinely valued, so the correct answer is a rate-card conversation rather than a cancellation.

The notice deadline is the sharpest object in the file. It is nine days away and it appears once, in the third paragraph of a covering email. Missing it is the easiest outcome available, it is invisible for twelve months, and it is the exact mechanism by which a contract renews itself at a seven percent uplift while everybody involved believes they are still deciding. The report therefore states, as a plain yes or no separate from every score, whether the date appeared anywhere in the candidate's output. That is a fact a hiring manager can act on without reading anything else.

The cards test the second thing that separates a strong office manager, which is applying a rule to a case it does not quite fit. The failure here is bimodal and both halves are common: quoting the policy back at somebody whose situation the policy did not anticipate, or waiving it informally and creating a precedent the next person will cite. The competent middle is to decide the case on its facts, say which part of the decision was discretionary, and record the ones that will be cited later. The gift card belongs with them rather than in a separate integrity section, because it is the same skill turned on the candidate themselves: a small courtesy, from an account manager they deal with monthly, offered while that supplier's renewal sits on their desk. Declining privately is not enough, and the anchor says so.

The disclosure card is the highest-risk thing this role does and it is scored under a constraint stated plainly: routing and discretion only. The fixture never says what the colleague's circumstances are, the candidate cannot ask, and no anchor anywhere reads their attitude to it. What is graded is whether they recognised that it was not theirs to hold, whether they told the person what would be passed on before passing it, whether they handled the practical part that genuinely was theirs, and whether the substance stayed out of every artefact where it does not belong. Both failures are scored 1 and both are floored — the one who promises to tell nobody and handles it informally, leaving no record and cutting the person off from whatever they may be entitled to, and the one who routes it correctly but tells the colleague afterwards or not at all. Neither can be offset by an excellent renewal analysis, and a score of 1 there is printed as a finding rather than averaged.

The last criterion is the one that distinguishes an office manager a company outgrows from one it comes to rely on. Every week offers the same choice: handle the request personally, which is faster today, or change the thing that generates it, which is slower today and compounds. The fifth card makes the aggregate visible — five people this month, eleven minutes each — and then asks for nothing. Nothing in the brief suggests writing anything down. Whether the candidate does is the observation, and it is the single most reliable predictor in the fixture of what this person will be worth in a year.

Thirty-eight minutes, allocated against a buyer who is usually a founder or a finance director hiring one person for a job whose title means five different things across five companies. The conventional screen asks whether the candidate has held the title before, which selects for having done a possibly unrelated version of it. Fourteen minutes of contracts and fourteen minutes of cases costs less than the second interview it replaces and answers a question that interview structurally cannot.

Sources

Every figure on this page is traceable. Where a claim could not be sourced it is stated qualitatively instead.

1. US Bureau of Labor Statistics, Occupational Outlook Handbook, Administrative Services and Facilities Managers, 2025, administrative services managers 279,000 jobs and facilities managers 165,900, 36,100 projected annual openings and 5 percent growth to 2035, https://www.bls.gov/ooh/management/administrative-services-managers.htm
2. US Bureau of Labor Statistics, Occupational Outlook Handbook, Secretaries and Administrative Assistants, 2025, https://www.bls.gov/ooh/office-and-administrative-support/secretaries-and-administrative-assistants.htm