1. Week 1, days 1 to 3: first pass on every application
On day 1 we open every application, not just the first ten that land before lunch. We read each one against a one-page criteria sheet we wrote before posting the role, because criteria written after reading CVs bend toward whoever sounded most polished. For a typical small-team role we expect 40 to 80 applications, and we give each first pass a fixed allowance: 4 minutes per application, which is illustrative, so the inbox closes in one focused block. At 60 applications that is 4 hours, and we split it across two mornings so attention does not sag after the first 90 minutes. We treat 15 to 20 of those as likely phone-screen candidates and mark the rest as a genuine no. The output by the end of day 3 is a shortlist with names and a single sentence per person explaining the advance or the rejection.
2. Week 1, days 4 to 5: thirty-minute phone screens
Day 4 and day 5 belong to phone screens. We book 30 minutes per candidate and never let them run into the next slot. We ask the same six questions in the same order for everyone shortlisted, because unstructured chats are the most reliable way to hire the person who talks best rather than the person who works best. For example, we open with "Walk me through one project you owned end to end" and follow with one role-specific probe and one logistics question about notice period and salary expectations. We stop at 30 minutes even when the call is going well, because an extra 15 minutes on a good call is usually an extra 15 minutes of noise. We expect 15 to 20 screens across the two days, and we allow 10 minutes after each call to write the screen note while memory is fresh; that is 40 minutes per candidate including the call, which is illustrative. By the end of day 5 we have a ranked group of 5 to 7 people moving to structured interviews, and everyone else gets a same-week decline email. The OPM structured interviews guidance we link at the end of this page is the reference we lean on when we argue that standardised questions are the fair and effective default, not a bureaucratic extra.
3. Week 2: structured interviews with the same questions
On day 8 and day 9 we run structured interviews, 45 minutes each, with a panel of two. We build the question set on day 6, the Friday before, so nobody improvises a favourite question at the last minute. We keep it to five questions and tie each one to a criterion: for a customer-facing role we ask for a time the candidate handled an angry customer, a time they simplified a process, a time they disagreed with a manager, a time they shipped under a hard deadline, and a time they asked for help. Every candidate hears the same five questions in the same order. Panel members write scores independently before the debrief, because discussion before scoring contaminates everyone's judgement toward whoever most resembles the first person to speak. We anchor the scale: 1 is a weak answer with no evidence, 3 is a solid answer with some specifics, 5 is a strong answer with clear ownership and a measured result. We tell every candidate from the start that this is one round, that we will decide within five working days, and that we will give a specific reason either way.
4. Work samples: when they help and when they exploit
Between the phone screens and the interview loop we decide whether a work sample earns its place in the schedule. Work samples help when the job produces an inspectable artifact and the exercise mirrors the real one: a two-hour editing test for a writer, a small spreadsheet cleanup for an analyst, a three-question debugging exercise for a support engineer. Work samples exploit when they ask for free labour on live problems, when they run over three hours, or when the role does not actually produce that artifact day to day. Our rule is simple. The exercise must be capped at 120 minutes, paid at the candidate's normal rate, and scored on a fixed rubric with four categories at 25 per cent each, because every add-on round adds cost; a two-hour paid sample times five candidates is two full working days of someone's time. If the role is coordination-heavy and has no clean artifact, we skip the sample and let the structured interview carry the weight. If we do run one, it sits on day 10 and we score it before the decision meeting, the same as everything else.
5. Decision meeting: scoring before discussion
Day 11 is the decision meeting, 60 minutes, three people, no more. Everyone arrives with their grid already filled in: phone screen score, structured interview scores per question, work sample scores where relevant. We calculate the weighted total before anyone speaks. In our default grid the structured interview carries 60 per cent, the phone screen 20 per cent, and the work sample 20 per cent; that is illustrative, and we adjust the weights before the first application arrives, never after a favourite candidate has emerged. The grid predicts the decision because the arithmetic is done in advance of the room. We discuss only the gaps: where two panellists are two or more points apart on the same answer, we revisit the notes and the candidate's exact words. Then we rank. If the top score is within 5 points of the second, we treat it as a tie and decide on the specific skill the team actually lacks this quarter. By day 12 we make the verbal offer, and by day 14 we send written confirmation. The two-week schedule produces one clear recommended hire and one genuine backup, which is the artifact you need before an offer goes out.
| Day range | Activity | Minutes per candidate | Output |
|---|---|---|---|
| Days 1 to 3 | First pass on every application | 4 | Shortlist of 15 to 20 phone-screen candidates |
| Days 4 to 5 | Thirty-minute phone screens | 40 including the write-up | Ranked group of 5 to 7 for structured interviews |
| Day 6 | Build the fixed five-question structured interview | 0 per candidate | One question set tied to criteria, no improvisation |
| Days 8 to 9 | Structured interviews with the same five questions | 45 plus 15 minutes of independent scoring | Independent scores per question from each panellist |
| Day 10 | Optional paid work sample | 120 capped | Rubric score on four weighted categories |
| Day 11 | Decision meeting, weighted totals pre-calculated | 60 for the meeting, not per candidate | One recommended hire and one backup |
Today, before you read another application, open your last job post and rewrite the criteria sheet so day 1 of the next cycle has a fixed scoring key waiting for it. That single change is what keeps the two-week schedule honest, and it is the first step in making candidate hiring predictable at small-team scale.
Sources
The substance of this page rests on the published material linked below. It is independent of this site and was read before writing.