Graduate Opportunities 2026: Administrative Officer Specialist opened on 28 September and closed at 3pm on 13 October. The booklet for vacancy 9005 gives two provisional windows: Stage 1, an online assessment, from 5 to 10 November 2026, and Stage 2, a capability-based interview and practical exercise, from 18 to 22 January 2027. Between application and the first window there are about three weeks. This guide is about those weeks.
What the booklet says, and what it does not
The booklet names four streams: AO (Energy, Environment and Climate), AO (Finance Policy), AO (Health Policy Analyst) and AO (Human Resources). It lists the selection methods publicjobs may use, from an online assessment questionnaire and online tests through shortlisting, interviews, a presentation or analysis exercise and a work sample. It sets out the AO Capability Framework in Appendix 1, with four dimensions, Building Future Readiness, Evidence Informed Delivery, Leading and Empowering, and Communicating and Collaborating, and says that specialist expertise “will be assessed throughout the process, and in particular, under the area of Leading with Specialist Insight”. On Stage 2 it says that familiarisation material and preparation guidelines will be sent to those invited.
What it does not say is what the Stage 1 test contains. That is normal: the familiarisation document arrives with the invitation. The best evidence of what to expect is the recent graduate AO and HEO competitions, which sat the same supplier’s tests, and the candidates who described them.
The test you should expect
AO 2025 and HEO 2026 both ran on the Sova Assessment platform, remotely proctored: one screen, one device, webcam check, a slow turn of the room and desk, calculator and blank paper shown to the camera. Three tests follow.
- Verbal reasoning. About 15 questions, passages in a policy or administrative register, each statement answered True, False or Cannot Say. No countdown on screen, but response time is recorded and forms part of the score.
- Numerical reasoning. About 15 questions on tables and charts, mostly percentages, shares, changes and projections. Again no visible clock, time recorded. HEO candidates in October described the difficulty climbing fast and the charts being thin on labels.
- Situational judgement. Ten scenarios, five actions each, 50 minutes. Each action is rated on a six-point scale from Highly Inappropriate to Highly Appropriate, independently of the others. Repeated ratings are allowed, there is no negative marking, and near-correct ratings earn partial credit.
The scoring rule is the detail to confirm when the familiarisation email arrives. In AO 2025 candidates reported that numerical had to be passed but did not count towards the ranking, which came from verbal and situational judgement. In HEO 2026 the booklet said all three scores were added. Either way the situational judgement test carries the most weight you can still move in two weeks, because the other two draw on skills a graduate already has. Our guide to Stage 1 across the grades has the regimes side by side.
What makes the AO scenarios different
The HEO test was described by people who sat it as the most Civil Service specific SJT they had seen. The AO version uses the same scale and, as far as anyone outside publicjobs can know, the same principles, written for a grade that analyses, briefs and advises rather than manages a team. Expect situations such as a briefing with a wrong figure that has already gone to the Minister’s office, a stakeholder who wants an early look at a submission, a dataset with personal data and a colleague who wants to run it through a public AI tool, a request from a TD’s office, a draft that an AP wants softened. The questions underneath are constant: what can an AO decide alone, who must be told and in what order, which actions breach a rule rather than a preference.
Three rules carry most of the marks. The action that quietly changes policy, waives a check, releases data or goes around your line manager is at the bottom of the scale, never in the middle. The action that fixes the problem inside your remit and tells the right person promptly is at the top. And the scale is absolute: if three of the five actions are good, three get a high rating. Candidates who spread their ratings to look balanced give away marks on every scenario. The situational judgement guide works through one scenario on the six-point scale.
Two weeks, in order
| Days | Do | Why in this order |
|---|---|---|
| 1 to 3 | Twenty to thirty six-point scenarios, one at a time, committing before you see the key and reading the explanation for every action you placed more than a step away from it. | By the third evening you can see your own habit: softening breaches, over-escalating, spreading ratings. That is the thing to fix, and the other tests do not need this many days. |
| 4 to 7 | Timed papers: ten scenarios in 50 minutes, no feedback until the end. One or two a day. | The real test gives no feedback between scenarios and the clock is on the whole paper. The habit has to hold without the explanations. |
| 8 to 10 | Verbal passages, aiming at about a minute per statement. For every miss, name the trap: scope shift, absolute word, motive leap, exception. | Almost all verbal errors are one habit: answering from what you know rather than what the passage says. Naming it is most of the cure. |
| 11 to 12 | Numerical tables with a clock running. Percentages of totals, change between periods, reverse percentages, ratios, projections. | On Sova your time is part of the score. The maths is school level; the speed is not, until practised. |
| 13 | Set-up: webcam, lighting, a clear desk, photo ID, calculator and paper to hand, phone out of the room, everyone in the house told. One scenario and one passage to stay warm. | Set-up failures cost candidates their slot. Nothing about the test itself improves on the last day. |
| Test day | Sit it early in the window if the booking allows. | Any technical problem then has days of window left to rebook into. |
After the window
Results come as a score and a position in the Order of Merit, and the position decides whether you are called in January and in which batch. Stage 2 is a capability-based interview and a practical exercise, with familiarisation material sent to those invited. Leave it until the result is in. The pay is set by Circular 22/2026: the AO scale from 1 June 2026 runs from €41,176 to €70,547, with a long service increment of €76,546, and the pay scales guide has the rest.
The Stage 1 trainer has an AO setting: fifty-six six-point scenarios built on the circulars, five timed papers of ten with no repeats, thirty verbal passages and numerical tables that change each time. The first scenario, passage and table are free. After that it is €25 for the month, which covers the January interview trainer too, or €75 for six months.



