Interviews: 4 ways operators run them on Pilla

#1 - Structured Q&A

Who it's for: Owner-operators and single-site managers who run every interview themselves. No HR function, no second interviewer, no formal scoring rubric.

Available on: Basic.

A structured Q&A interview is the leanest defensible hiring process: the same five questions asked in the same order to every candidate, notes captured under each one, a single overall fit rating, and a recorded decision. The Chartered Institute of Personnel and Development found that structured interviews are roughly twice as predictive of on-the-job performance as unstructured ones. Most single-site operators don't run unstructured interviews because they want to; they run them because nobody has handed them a script.

A single-site café owner picks up their phone in the back office, opens the canvas, types in the candidate's name and the role they applied for, then walks the candidate through the five questions in the same order they've used for the last six hires. Notes go in under each question. Fit rating at the end. Decision recorded before they walk back to the floor. The whole interview takes 25 to 35 minutes and produces a record that a recruitment dispute, a probation conversation, or a re-hire decision can reference six months later.

What makes it work as the lean end of the spine: the five questions live in the canvas as a checklist. The operator doesn't have to remember them. The rating scale is anchored at 5 = would hire, 3 = would consider, 1 = no, with the explicit instruction to avoid 3 unless they genuinely mean it. That single nudge cuts the "everyone's a maybe" failure mode that wrecks single-site hiring.

When to upgrade from this version:

1. You start delegating interviews to other managers and need them to ask the same questions the same way
2. A second-round panel format is creeping in and notes aren't comparable across interviewers
3. You hire 20+ people a year and want defensible records against future disputes
4. A regulator (CQC, OFSTED) starts asking how you ensure hiring decisions are fair

#2 - Interview with guidance

Who it's for: Operators with 2-5 sites who delegate hiring to local managers and want consistency in how interviews are run, not just what's asked.

Available on: Standard.

An interview with guidance is the structured Q&A plus rich-text panels alongside the four sections where new interviewers most often go wrong: question delivery, note-taking, rating honestly, and deciding without delay. The guidance steers behaviour the way a senior manager would if they were sitting in. Roughly 1 in 3 hiring decisions across hospitality and retail are made by someone who hasn't been formally trained as an interviewer; written guidance on the canvas closes that gap without scheduling an in-person training session.

A three-site pub group is the canonical operator. The owner hires once or twice a year; their three site managers hire a dozen times each. Without guidance the three sites end up running three different interviews despite using the same five questions. The "Stick to the script" panel above the question checklist removes the temptation to riff. The "Capture what they said, not what you thought" panel above the notes field stops the rating bias that creeps in when notes are already a verdict. The "Avoid 'maybe'" panel above the rating cuts the centre-bias that produces a bench of unhirable threes. The "Decide now" panel above the decision field stops the sleep-on-it stall that turns hires into ghosts.

The build is the V1 chain with four template_rich_text nodes inserted before Questions, Notes, Overall fit, and Decision. Each panel is short, two or three lines, written as the manager would be told if they were being briefed in person. The chain still runs top to bottom on a phone in 25-35 minutes; the panels add reading time but cut interpretation drift.

What this version unlocks:

1. Three different managers run the same interview the same way
2. Notes capture content, not interpretation
3. Ratings cluster appropriately at the extremes rather than the centre
4. Decisions get made on the day, not deferred indefinitely

When to upgrade from this version: when a regulator or HR function wants evidence that interviews are run in a specific sequence (Interview #3), or when a hire dispute requires a complete audit trail (Interview #4).

#3 - Interview with locked sequence

Who it's for: Care homes, nurseries, hospital trusts, and HR-led operations that need to demonstrate a consistent interview process for every candidate.

Available on: Standard.

An interview with locked sequence is the guided interview with flow_kind: 'gated' edges between consecutive check_element rows, so the interviewer can't sign off the decision before completing the questions, taking notes, and recording the rating. The CQC's well-led key question explicitly assesses recruitment process consistency; ACAS guidance on protected-characteristic recruitment disputes recommends a documented, repeatable interview sequence as the strongest defence. Gated edges turn that recommendation into a feature of the canvas, not a hope.

A 12-site care provider with regional managers running roughly 8 hires a month per region is the canonical operator. Before locked sequence the regional managers would sometimes complete the rating before fully writing up the notes; in a tribunal those out-of-order timestamps read as the rating driving the notes rather than the other way around. With gated edges the timestamp story is unambiguous: candidate name, then role, then questions worked through, then notes written, then rating given, then decision recorded. Each timestamp comes after the previous.

The technical mechanism is the same V2 chain (guidance panels alongside the check elements) but with the edges between consecutive check_elements set to gated rather than open. Rich-text panels still flow open because guidance doesn't gate; it lives in the background while the worker reads it. Standard plan unlocks gated edges; Basic keeps everything open.

This version's specific value:

1. Every interview produces timestamps in the right order, defensible against discrimination claims
2. The "decide before writing up" failure mode is structurally impossible
3. Regional managers run hiring the same way the head office process specifies
4. The CQC inspector reading recruitment records sees consistent process, not consistent paperwork

When to upgrade from this version: when the interview decision needs to be backed by physical evidence (annotated CV, interviewer signature) for audit purposes (Interview #4).

#4 - Interview with evidence

Who it's for: Multi-site chains, healthcare groups, retail estates with hiring volume in the hundreds per year and the corresponding exposure to unfair-dismissal claims, discrimination disputes, or DBS/safer-recruitment audits.

Available on: Standard.

Interview #4 builds on the locked sequence by adding two media elements at the end: an annotated_photo of the candidate's CV with the interviewer's highlights and notes drawn directly on the image, and a signature from the interviewer confirming the decision. The CV photo captures what an interviewer noticed at the moment they read it (a gap, a strong reference, a missing date) rather than what they remembered later. The signature locks the rating and decision against later second-guessing.

A 40-venue casual dining group running 200+ hires a year is the canonical operator. Their HR director needs to be able to defend any individual decision against a tribunal challenge. Pre-evidence, that meant 8 separate paper trails (job description, interview notes, decision rationale, references, contract, DBS, induction, probation). Post-Pilla-with-evidence, the interview canvas alone holds the conversation, the rating, the annotated CV, and the interviewer's signature, all timestamped to the interview itself. The remaining HR records sit alongside but the interview moment is captured once and unambiguously.

The canvas is the V3 sequenced chain plus two new top-level chain nodes at the end: the annotated_photo (with instruction to circle strengths in one colour and gaps in another), then the signature. Both edges into the evidence appendix are gated, so the interviewer can't sign off without first uploading the annotated CV. The signature unlocks only when the CV photo is captured.

What the evidence actually captures, and why each:

1. Annotated CV photo: the interviewer's attention at the moment they read the document, not their reconstruction of it weeks later
2. Interviewer signature: a clear commitment to the rating and decision, defensible against "I never actually signed off on that hire" later
3. Combined with the rating and notes: a complete audit trail that ACAS treats as best-practice evidence of fair process

When to upgrade from this version: when the work is too involved for a single interviewer to remember context, and you want AI to brief the interviewer on the candidate's CV before the conversation, gate whether the interview proceeds based on availability, or post the decision automatically to the HR system. Those variations are coming in the next post update.

How to pick the right version

Three questions to narrow it down:

• Are you running every interview yourself, or delegating to other managers? Yourself: Interview #1. Delegating: Interview #2 onwards.
• Does a regulator (CQC, OFSTED) or HR function check that interviews are run in a specific order? No: Interview #1 or #2. Yes: Interview #3 onwards.
• Do you need defensible records against future disputes or tribunals? Records of conversation: #1, #2, or #3. Evidence of the conversation: Interview #4.

Related reading

• Onboarding - what happens after the offer is accepted
• Performance reviews - the structured review that mirrors a structured interview
• One-to-ones - the regular 1-to-1 cadence that keeps the new hire on track
• Recruitment use case - every Pilla canvas around hiring in one place

Frequently asked questions

How long should an interview take?

Between 25 and 45 minutes, depending on the version and the role. A single-manager structured Q&A for a frontline role runs 25-35 minutes. An evidence-backed interview for a regulated role (care, finance, safeguarding) runs 35-45 minutes including the time to annotate the CV and sign off. Longer than 60 minutes is usually a panel format, which is a different canvas.

Why structured interviews over unstructured?

The Chartered Institute of Personnel and Development's research shows structured interviews are roughly twice as predictive of on-the-job performance as unstructured. Unstructured interviews mostly measure interviewer-candidate rapport, which is a poor proxy for whether someone will do the job well. Structured interviews ask every candidate the same questions and rate them on the same scale, so the comparison is real.

Do I need an annotated CV photo for an ACAS-compliant interview?

No. ACAS guidance recommends a documented interview process with consistent questions and rating, which Interview #3 covers. The annotated CV in Interview #4 is a defensive practice taken by chains and groups with high hiring volume and corresponding tribunal exposure. If you hire under 50 people a year, Interview #3 is sufficient.

Can a candidate ask to see their interview notes?

Yes, under UK GDPR. A candidate can submit a Subject Access Request and you must provide their interview notes within one calendar month. Notes captured in Pilla are timestamped and exportable; the structured rating sits alongside, which makes the SAR response straightforward. This is one reason structured notes beat shorthand scribbles on a paper sheet.

What's the difference between a gated edge and an open edge for interviews?

An open edge means every section of the interview canvas is available immediately; the interviewer can take notes before asking questions if they want to. A gated edge means the target section is locked until the source section is complete; the interviewer can't record a rating before the notes are written, and can't sign off before the rating exists. Interview #3 and #4 use gated edges to enforce sequence integrity.

How does this work for panel interviews?

The canvases above are designed for single-interviewer flow. For a panel, two patterns work: one interviewer captures the canonical record while others contribute notes verbally, or each panellist runs the canvas independently and the records get compared after. The structured rating scale in the canvas is what makes either approach defensible; panel disagreement on a rating is the signal to discuss further before deciding.

Where to go next

A structured interview is roughly twice as predictive as an unstructured one, but the gap only opens up if the structure is actually used. Operators running Interview #3 or #4 self-enforce that structure through gated edges and evidence capture rather than relying on the interviewer to remember.

Five more variations are coming in the next refresh that bring AI into the interview: Poppi briefing the interviewer on the candidate's CV before the conversation, gating whether the interview proceeds, posting the decision to an ATS, deciding whether to escalate borderline candidates, and routing the offer process by rating tier. Those need more review time and will land separately.

→ Build your own interview canvas on Pilla. Basic plan unlocks Interview #1 today.