Observation is the assessment method examiners scrutinise hardest and learners fear most — get the structure right and it stops being stressful for either side.
TL;DR
To conduct an observation as an assessor, you plan against the assessment criteria beforehand, watch without interrupting, record evidence in real time, and give feedback that links directly to the standard. Verdict: a planned, criteria-mapped observation with immediate feedback is the only version that holds up at internal quality assurance (IQA) or external quality assurance (EQA) sampling in 2026. Assessors who wing it lose evidence on cross-examination questions and lose learner trust. This guide covers what you need before you walk in, the step-by-step process, and where most CAVA-qualified assessors trip up.
Why this matters
Observation is direct evidence of occupational competence — it's the one method that can't be faked with a written statement. Ofqual-regulated frameworks like the Level 3 Certificate in Assessing Vocational Achievement treat it as a core method precisely because a learner either performs the task to standard or doesn't.
Bad observations create two problems. First, weak or missing notes get flagged at IQA sampling and the assessment gets invalidated, forcing a re-observation. Second, learners who feel ambushed or judged unfairly disengage, which shows up later as withdrawal rates. Getting the process right the first time protects both the qualification's integrity and the learner relationship.
What you'll need
- The unit specification and assessment criteria for the qualification being assessed
- A completed assessment plan agreed with the learner in advance
- An observation checklist or record form mapped to the criteria
- 30-90 minutes of uninterrupted time, depending on the task
- A quiet method for recording notes (paper or tablet) that doesn't distract the learner
- Your CAVA qualification or equivalent assessor award, current and on file
The steps
1. Agree the plan with the learner beforehand
Never observe cold. Confirm the date, time, task, and which criteria you're assessing against at least a week ahead where possible.
This matters because CAVA Unit 1 requires assessment to be planned, not sprung on learners. An observation without a documented plan is one of the first things an IQA checks in a sample, and it's an easy fail to avoid.
Common mistake: turning up unannounced to see them in action. That's surveillance, not assessment, and it won't stand up to scrutiny.
2. Confirm the criteria you're checking against
Before you watch anything, write out exactly which performance criteria the observation will cover — not the whole unit, just what's realistically demonstrable in this session.
This focuses your attention during the observation itself. Trying to assess 12 criteria in one 20-minute task means you'll miss evidence for most of them.
Expected outcome: a short checklist, criteria-coded, that you can tick or annotate live.
3. Brief the learner on the day
Spend two minutes reminding the learner what you're observing and confirming they're ready. Ask if there's anything unusual about today's conditions — short-staffed shift, different equipment, a client who's had a bad morning.
This matters because context changes what competent performance looks like. A care worker managing a resident's unexpected distress is still demonstrating competence even if the task doesn't go as scripted.
Common mistake: skipping the brief and then penalising the learner for deviating from a plan they didn't know you were checking against.
4. Observe without intervening
Watch the full task from start to finish. Take notes as it happens — what was said, what was done, in what order, with what outcome. Don't prompt, correct, or step in unless there's a genuine safeguarding or safety risk.
This is the core evidence-gathering moment and it's why observation outranks a written statement: you're watching unassisted, real-world performance. If you intervene, you've turned the observation into training, and the evidence is compromised.
Specific instruction: note timestamps for key actions (14:32 — checked ID before administering medication) so your record reads as contemporaneous, not reconstructed from memory afterward.
Expected outcome: a written record with enough specific detail that a third party — an IQA or EQA — could read it and understand what happened without being there.
5. Record evidence against each criterion immediately
Within the same session, map what you observed to the specific criteria on your checklist. Don't wait until the next day — detail degrades fast, and vague notes (did it well) are the single most common reason observation records get challenged.
This matters because your record is the permanent evidence file. If it's thin, the assessment decision is unsupportable regardless of how competent the learner actually was.
Common mistake: writing a general narrative summary instead of criterion-by-criterion evidence. IQAs need to trace each judgement back to a specific criterion.
6. Give feedback the same day
Deliver feedback to the learner as soon as practical after the observation — ideally within the hour, never more than a working day later. Confirm what was competent, what wasn't yet meeting the standard, and what's needed to close any gap.
This matters for learner development and for the assessment decision itself. Feedback should reference the criteria directly (you met 3.2 and 3.4 but I didn't see evidence for 3.3) rather than general praise or criticism.
Expected outcome: the learner leaves knowing exactly where they stand and what happens next — a resit, additional evidence, or sign-off.
7. File the record and cross-reference other methods
Add the completed observation record to the learner's portfolio and note which other assessment methods (questioning, professional discussion, work products) still need to fill any gaps.
This matters because observation rarely covers a whole unit alone — it's typically combined with 2-3 other methods, a principle covered in detail in the guide to CAVA Unit 2 on assessing occupational competence.
Common mistake: treating one observation as sufficient evidence for an entire unit when the criteria clearly need supplementary methods.
Troubleshooting
The learner seems nervous and underperforms. Reschedule if it's clearly affecting genuine competence — a one-off nerves-driven dip isn't representative, and forcing a fail here damages trust for no evidential gain.
You can't observe every criterion in one session. Split the observation across two sessions or supplement with professional discussion afterward. Partial evidence plus a follow-up method is standard practice, not a shortcut.
The workplace setting is unpredictable (care homes, classrooms, live client work). Build flexibility into the assessment plan — note that the observation window may shift by a day or two given the setting, and record what actually happened rather than what was scheduled.
IQA feedback says your notes are too vague. Go back to criterion-by-criterion phrasing with specific actions, times, and outcomes. Demonstrated good communication is not evidence; explained the procedure in plain terms and checked the client understood before proceeding is.
The learner disputes your judgement. Show them the specific criteria and the specific gap in the record. If the evidence genuinely supports a different call, revisit it — assessment isn't about defending your first instinct, it's about accuracy.
You're assessing a task that rarely occurs naturally (e.g. emergency procedures). Use simulation where the standard allows it, and record clearly that it was simulated, not naturally occurring, since IQAs weight these differently.
Tools and resources
- Roles and responsibilities of a vocational assessor — the full scope of the assessor role beyond observation
- How to prepare your CAVA portfolio of evidence — filing and organising observation records alongside other evidence
- A criterion-mapped observation checklist template from your awarding organisation
- Your qualification's unit specification document, kept to hand during every observation
FAQ
What's the best way to record an observation as an assessor?
Criterion-by-criterion notes taken in real time, with specific actions and timestamps, beat a general narrative summary every time — vague notes are the most common reason observation evidence fails IQA sampling in 2026.
How long should an observation last?
Long enough to demonstrate the criteria naturally, usually 20-90 minutes depending on the task; padding a short task to fill time doesn't add evidential value.
Is observation better than a witness statement?
Observation is stronger because it's direct, unassisted evidence gathered by a qualified assessor, whereas a witness statement relies on a third party's account and typically needs supporting evidence alongside it.
Do I need to warn the learner before observing?
Yes — assessment must be planned and agreed in advance under CAVA principles; unannounced observation isn't valid assessment practice and won't survive IQA scrutiny.
Can one observation cover a whole qualification unit?
Rarely — most units need observation combined with 2-3 other methods like questioning or professional discussion to cover every criterion fully.
What happens if a learner doesn't meet the standard during observation?
You record exactly what was and wasn't demonstrated, give specific feedback on the gap, and agree a plan for further evidence or a repeat observation — it's not an automatic fail of the whole unit.
How much detail does an observation record need?
Enough that someone who wasn't there — an IQA or EQA — can understand exactly what happened and trace each judgement to a specific criterion; a one-line summary isn't sufficient.
Do I need a CAVA qualification to observe learners?
Yes, in most regulated frameworks you need a recognised assessor qualification such as the CAVA award before your observation judgements count as valid assessment decisions.
One last thing
The assessors who get sampled cleanly aren't the ones with the most experience — they're the ones whose notes read like a transcript. If an IQA can reconstruct exactly what happened from your record alone, without asking you a single follow-up question, you've done the job right.
