# Global Quality Seal: Lead Reviewer Training *Applying VRITE consistently in desk review and interview.* ## About this course This course is for ICoBC Lead Reviewers (N2). A Lead Reviewer plans a review, leads a small review team, tests the evidence an issuer submits, and signs a report that the Quality Network can rely on. The course is aligned to the ICoBC Credentialing Skills Framework (ICSF), skill QUAL "Quality assurance and accreditation" (VRITE pillar: Rigor), level L4 "Assure", stage A "Ability". At this level the skill is to design and lead internal quality assurance and to conduct independent reviews of other issuers. At the Ability stage you practise that capability on realistic tasks in a controlled setting. **Prerequisites:** ICB-F09 Quality in Credentialing: How Trust Is Built, and ICB-R04 Independence and Conflicts of Interest in the Quality Network. **Workload:** 16 hours notional learning, in a blended cohort. This file covers the self-paced part. Live and workshop elements appear as facilitator notes inside each module. **Learning outcomes.** When you finish, you can: 1. Apply VRITE criteria and indicators consistently to issuer evidence. 2. Conduct desk reviews and review interviews. 3. Write findings and recommendations that are evidence-based and fair. 4. Maintain independence and confidentiality. **How it is assessed.** Each module has a quiz with feedback and a practice activity. The assessed task is a full review of a case file with a written report, scored by a person against the reviewer rubric in the section "Assessment and badge". You also complete one shadow review. A final knowledge check supports the task. Nothing in this file grades your report. **Badge.** "Global Quality Seal: Lead Reviewer Training (Ability, L4)", alignment code ICSF-QUAL-L4-A. Badge evidence: your case report and your shadow review record. The badge records that you met the assessment criteria for this course. It does not prove competence in any job. **Quality focus:** core VRITE application. **AI focus:** use and disclosure of AI in review work. | Outcome | Module | How it is assessed | |---|---|---| | 1. Apply VRITE criteria and indicators consistently to issuer evidence | Criteria and indicators | Module quiz and practice activity; criteria application in the case report | | 1. Apply VRITE criteria and indicators consistently to issuer evidence | Desk review | Module quiz and practice activity; evidence table in the case report | | 2. Conduct desk reviews and review interviews | Desk review | Module quiz and practice activity; case report and shadow review record | | 2. Conduct desk reviews and review interviews | Interview technique | Module quiz and practice activity; shadow review record | | 3. Write findings and recommendations that are evidence-based and fair | Writing findings | Module quiz and practice activity; findings section of the case report | | 4. Maintain independence and confidentiality | Ethics | Module quiz and practice activity; declarations in the case report and shadow review record | | All outcomes, orientation | The Seal and the review process | Module quiz and practice activity; final knowledge check | > **Status of the Seal.** The Global Quality Seal and the ICoBC VRITE criteria are **[Emerging]**. Indicators are not yet published. Internal documents use different expansions of the acronym VRITE. Pilots are ongoing. This course therefore teaches the review method and uses illustrative indicators that are labelled as such. Status as of November 2026; check before publishing. ## The Seal and the review process In this module you learn what a review is for, who does what in it, and the order of its stages. Later modules build the skills for each stage. ### What the Seal is and what a review does The Global Quality Seal is a quality mark that ICoBC is developing for issuers of credentials. A review is the structured check that supports a decision about whether an issuer meets the VRITE criteria. VRITE is the ICoBC framework of quality pillars, and this course works with the Rigor pillar as it applies to reviewing. Because the Seal and its indicators are **[Emerging]**, you will meet three facts in practice: - Indicators are not yet published, so reviewers work from the criteria and from guidance given for each pilot. - Internal documents use different expansions of the acronym VRITE. Use the expansion given in your engagement brief and do not substitute your own. - Pilots are ongoing, so procedures may change between engagements. Record the version of every document you use. A review has one purpose: to give the decision-making body a fair, evidence-based picture. The review team does not decide the outcome. It reports what the evidence shows against the criteria. ### Roles in a review Each review has defined roles. Names of bodies and committees vary by engagement, so confirm them in your brief. - **The issuer** is the organisation under review. It submits an application and evidence. - **The Lead Reviewer** plans the review, assigns work, leads interviews, checks the quality of the team's notes and signs the report. - **Reviewers** (where the team has more than one) examine evidence and take part in interviews. - **The ICoBC review coordinator** handles logistics, documents and communication with the issuer. - **The decision-making body** receives the report and decides. Reviewers do not. A Lead Reviewer carries two duties at once: to the issuer, who deserves a fair and well-run process, and to the Quality Network, which relies on consistent reviews. ### The review process from application to decision The stages run in this order: 1. **Application.** The issuer applies and submits its evidence. 2. **Conflict check and assignment.** The Lead Reviewer and team declare any conflicts. The coordinator confirms the team. 3. **Planning.** The Lead Reviewer reads the brief, lists what must be tested and sets the schedule. 4. **Desk review.** The team studies the documents and records what each document shows. 5. **Interviews.** The team tests and clarifies the desk findings with the people who run the work. 6. **Findings.** The team agrees findings and recommendations. 7. **Report.** The Lead Reviewer writes and signs the report. 8. **Issuer factual accuracy check.** The issuer may correct errors of fact. 9. **Decision.** The decision-making body decides. ### How the stages link Each stage feeds the next. Planning decides what the desk review must test. The desk review produces questions for the interview. The interview confirms or corrects what the documents seemed to show. Findings rest on both. The report presents findings so the decision-making body can follow the trail from evidence to conclusion. **Worked example: A first reading of a review brief** A Lead Reviewer, Amara, receives a brief from the ICoBC review coordinator for an engagement with a partner training provider (the issuer). The brief lists the criteria, the document set, a two-week desk review window and one half-day interview slot. Amara reads it in this order: 1. She checks the version number of the criteria document and notes it in her log. 2. She notes that no published indicators exist, so she reads the pilot guidance for how criteria should be interpreted. 3. She lists three questions for the coordinator: which expansion of VRITE applies, whether a factual accuracy check is part of this engagement, and who the decision-making body is. 4. She sends her conflict declaration before she opens any issuer document. Amara did not open the evidence first. She fixed the rules of the review before she began, so that every later judgement can be traced to a stated basis. **Practice activity: Plan the first week of a review** You are Lead Reviewer for a pilot review of a partner issuer called Northfield Learning Partners (an invented name for this exercise). Write a first-week plan of no more than 250 words that covers: - the documents you will read first and why, - the three questions you will send to the coordinator, - your conflict declaration step and when it happens, - how you will log document versions. Compare your plan with a peer in your cohort. Each of you names one thing the other plan lacks.

Facilitator notes

**Live element (60 minutes).** Run a plenary walk through the process graphic. Ask participants to name where each pilot variation could change the process (for example, the factual accuracy check). Then run the practice activity in pairs for 20 minutes, with 15 minutes of feedback. Collect two or three plans for later use as examples, with permission. **Watch for:** plans that start with the evidence and skip the brief and the conflict declaration.

**Question 1.** What is the role of the review team in relation to the decision on the Seal? [( )] It decides whether the issuer receives the Seal. [(X)] It reports what the evidence shows against the criteria so the decision-making body can decide. [( )] It negotiates the outcome with the issuer. [( )] It advises the issuer on how to pass. [[?]] Think about who receives the report. **** The review team gives a fair, evidence-based picture. The decision-making body decides. Reviewers who decide, negotiate or coach would damage independence. **** **Question 2.** Indicators for the VRITE criteria are not yet published. What should a Lead Reviewer do? [( )] Write personal indicators and apply them to all issuers. [( )] Wait until indicators are published before starting any review. [(X)] Work from the criteria and the pilot guidance, and record which versions were used. [( )] Ask the issuer which indicators it prefers. [[?]] The method must stay consistent and traceable. **** The Seal is Emerging. Reviewers use the criteria and guidance supplied for the pilot and log document versions. Personal indicators would make reviews inconsistent, and the issuer should not choose its own yardstick. **** **Question 3.** Internal documents use different expansions of the acronym VRITE. What is the best action? [( )] Choose the expansion you find clearest. [( )] Use the expansion from your last engagement. [( )] Avoid the acronym in the report. [(X)] Ask the coordinator which expansion applies and use it consistently in the report. [[?]] Consistency inside one report matters. **** Confirming the expansion with the coordinator keeps the report consistent with the engagement brief. Reviewers do not substitute their own expansion. **** **Question 4.** When should a Lead Reviewer send a conflict declaration? [( )] After the interviews, once the team knows the issuer well. [( )] Only if the issuer asks for one. [(X)] Before opening any issuer document, and update it if circumstances change. [( )] At the end, with the signed report. [[?]] A conflict can affect how you read the first document. **** Declaring before you see the evidence lets the coordinator reassign the work if needed. Circumstances can change, so the declaration is updated during the review. **** ## Criteria and indicators This module teaches you to apply criteria and indicators the same way every time, to every issuer. Consistency is the heart of the Rigor pillar for reviewers. ### From criterion to indicator to evidence A **criterion** states what good looks like. An **indicator** is an observable sign that the criterion is met. **Evidence** is the material you can examine: a document, a record, a data extract or something said in an interview. The chain runs in one direction: criterion, then indicator, then evidence, then judgement. If you start from evidence and look for a criterion that fits, you will find what you expect to find. Because VRITE indicators are **[Emerging]**, the examples in this course use **illustrative indicators**. They show the method. They are not the official indicators and must not be quoted as such in a real review. ### Illustrative indicator example Take an invented criterion: "The issuer defines the learning outcomes for each credential." Illustrative indicators might be: - Each credential has a written statement of outcomes. - The statement uses verbs that can be observed or measured. - The statement was approved by a named body and carries a version date. A reviewer looks for evidence against each indicator. Each indicator is judged on its own. A credential can meet the first and fail the third. ### Consistency across cases Consistency means that two reviewers looking at the same evidence reach the same judgement, and that you reach the same judgement for similar evidence from different issuers. Four habits help: 1. **Write the indicator before you read the evidence.** Keep the wording in front of you. 2. **Use a fixed scale.** Use the rating scale from the engagement brief. Do not invent half points. 3. **Record the evidence reference.** Note the document name, version and page, or the interview date and speaker role. 4. **Calibrate.** Before the review, the team reviews one sample item together and compares judgements. Differences are discussed and the shared reading is written down. ### Common threats to consistency - **Halo effect:** a strong first impression colours later judgements. - **Anchoring:** a first rating pulls the next ones towards it. - **Leniency or severity drift:** standards loosen or tighten as fatigue grows. - **Confirmation:** you seek only evidence that supports your first view. A Lead Reviewer names these threats to the team at the start and checks for them when reading draft ratings. ### Use of AI in applying criteria AI tools may be offered to help summarise documents or sort evidence. In review work, treat AI output as unverified. You remain responsible for every judgement. Do not paste confidential issuer material into any tool unless your engagement terms allow it and the tool's data handling has been approved. Disclose any AI use in the report, as module 6 explains. **Worked example: Calibrating a rating on outcomes statements** Two reviewers, Jonas and Priya, examine a sample credential description from a partner training provider in a pilot engagement. The illustrative indicator reads: "The statement uses verbs that can be observed or measured." The statement says: "Learners will understand data protection and appreciate its importance." - Jonas rates the indicator as met, because the topic is clear. - Priya rates it as not met, because "understand" and "appreciate" cannot be observed. They discuss. The indicator concerns the verbs. The topic's clarity belongs to a different indicator. They agree on "not met" for this indicator and write the reasoning in the calibration note: "Verbs such as understand and appreciate are not observable; statements with these verbs rate as not met unless an assessment criterion makes them observable." Both use this reading from then on. **Practice activity: Rate and justify** Below are three invented outcomes statements. Using the illustrative indicator "The statement uses verbs that can be observed or measured", rate each as met or not met and write one sentence of justification that cites the wording. 1. "Participants will be familiar with the safety rules." 2. "Participants will list the five stages of the safety procedure and carry out the first two." 3. "Participants will gain a deep awareness of safety culture." Then compare with a peer. Where you differ, write a calibration note that would settle the difference.

Facilitator notes

**Live element (90 minutes).** Run a calibration workshop. Give each participant three short invented evidence excerpts and the same illustrative indicator. Collect ratings anonymously, display the spread and discuss the outliers. Ask for a written calibration note for each disagreement. Close by naming the biases from the module and asking each participant which one they are most prone to. **Watch for:** participants who rate the quality of the writing instead of the indicator.

**Question 1.** What is the correct order in the chain of judgement? [( )] Evidence, indicator, criterion, judgement. [( )] Judgement, evidence, criterion, indicator. [(X)] Criterion, indicator, evidence, judgement. [( )] Indicator, judgement, criterion, evidence. [[?]] Start from what good looks like. **** The criterion states what good looks like, the indicator makes it observable, the evidence is what you examine, and the judgement follows. Starting from the evidence invites confirmation bias. **** **Question 2.** Why does this course use illustrative indicators? [( )] The official indicators are secret. [(X)] Official indicators are not yet published, so illustrative ones demonstrate the method without being quoted as official. [( )] Illustrative indicators are better than official ones. [( )] Reviewers are expected to write their own indicators. [[?]] Think about the status of the Seal. **** The VRITE indicators are Emerging and unpublished. Illustrative indicators teach the method and must not be presented as official in a real review. **** **Question 3.** Two reviewers rate the same evidence differently. What is the best response? [( )] The Lead Reviewer chooses the higher rating to be fair to the issuer. [( )] Average the two ratings. [(X)] Discuss against the indicator wording, agree a reading and record it in a calibration note. [( )] Ask the issuer which rating is correct. [[?]] The aim is a shared reading that can be repeated. **** A calibration note records the shared reading so that future judgements are consistent. Averaging or choosing the higher rating hides the disagreement without resolving it. **** **Question 4.** A reviewer's first rating for an issuer is high, and later ratings tend to follow it closely. Which threat to consistency is this? [( )] Leniency drift. [(X)] Anchoring. [( )] Calibration. [( )] Triangulation. [[?]] A first value pulls the later ones. **** Anchoring happens when an early judgement pulls later judgements towards it. The remedy is to rate each indicator on its own evidence and to check draft ratings for clustering. **** ## Desk review The desk review is where most of the evidence is examined. A good desk review is planned, systematic and recorded so that another reviewer could follow it. ### Plan the desk review Before reading, write a short desk review plan: - the criteria and indicators in scope, - the document list, with version and date for each, - what each document could show and what it cannot show, - the time you will allow per criterion, - the questions you expect to carry into the interview. A document shows that something was written. It does not by itself show that it is used. This gap between documented and practised is what the interview later tests. ### Read documents in a fixed way For each document, make a short record: 1. **Identify it:** title, version, date, author or owner, and status (draft or approved). 2. **Note what it says** that is relevant to an indicator, with page or section. 3. **Note what it does not say** where you would expect it to. 4. **Mark open questions.** Keep observation and interpretation apart. "The policy was last revised in 2021" is an observation. "The policy is out of date" is an interpretation that needs a stated standard to compare against. ### Test evidence quality Ask four questions of each piece of evidence: - **Authentic:** is it what it claims to be, from whom it claims? - **Current:** is it in force during the review period? - **Relevant:** does it speak to the indicator? - **Sufficient:** is there enough to support a judgement, or only one example? A single approved policy may be enough to show that a policy exists. It is not enough to show that the policy is applied. For application you need records such as meeting minutes, samples of completed forms or data extracts. ### Triangulation Where possible, support a judgement with at least two kinds of evidence, for example a document and an interview, or a document and a record sample. This is called triangulation. If sources disagree, do not choose the one you prefer. Record the conflict and take it to the interview. ### Sampling When an issuer has many credentials or many records, you cannot read all of them. State how you chose the sample, and why. A sample should cover the range of credentials, not only the best ones. Do not state a finding about the whole population from a small sample unless you say how the sample was chosen and how large it was. ### Confidentiality in the desk review Issuer documents may contain personal data about learners and staff. Store them only in the location named in the engagement terms. Do not copy them to personal devices. If you use an AI tool to summarise any issuer document, you need prior permission under the engagement terms, and you must disclose it in the report. **Worked example: Document says yes, record sample says not quite** Lead Reviewer Tomasz examines an invented partner engagement: a vocational school (the issuer) that issues a digital certificate. The illustrative indicator reads: "Assessment decisions are checked by a second person." - **Document:** the assessment policy, version 3, approved, says that all final assessments are second-marked. - **Record sample:** Tomasz asks for ten completed assessment files, chosen by him from a list of 120 (illustrative figure). Seven show a second marker's signature. Three show none. - **What he records:** "Policy v3 states second marking for all final assessments (section 4.2). In a sample of ten files chosen by the reviewer from a list of 120, seven showed second marking and three did not. The sample is small and does not allow a statement about the whole population." - **Open question for the interview:** "How is second marking recorded, and what happens when it is skipped?" Tomasz did not write "the school does not second-mark". He wrote what he saw, said how the sample was chosen, and carried a question forward. **Practice activity: Build an evidence table** You review an invented issuer, Lakeside Skills Association, against the illustrative indicator "Complaints are logged and answered within a stated time." You have: - A complaints policy (version 2) that states a 15 working day response time. - A complaints log extract of eight entries (illustrative): five show a response date within 15 working days, two show no response date, one shows 22 working days. Create a table with the columns: Evidence item, What it shows, Quality check (authentic, current, relevant, sufficient), Open question. Then write two sentences that state your provisional finding without overstating it.

Facilitator notes

**Live element (75 minutes).** Provide a mock document set of four short documents with planted inconsistencies (a draft marked as approved, a record outside the review period, a policy that contradicts a log). Participants work in pairs to find them and complete an evidence table. In the debrief, ask how each planted issue would change the interview plan. **Watch for:** participants who write conclusions in the observation column.

**Question 1.** A policy says all assessments are second-marked. What does the policy alone show? [( )] That second marking takes place in every case. [(X)] That a rule exists; records are needed to show whether it is applied. [( )] That the issuer meets the indicator in full. [( )] That the issuer has no problems with assessment. [[?]] Documented and practised are different. **** A policy is evidence of intent. Application needs records such as signed files or logs. The interview then tests how the rule works in practice. **** **Question 2.** Which sentence is an observation rather than an interpretation? [( )] The policy is badly out of date. [( )] The issuer does not care about complaints. [(X)] The policy on the file is version 2, dated 14 March 2021 (illustrative). [( )] The issuer's governance is weak. [[?]] An observation can be checked by anyone with the document. **** An observation records what the document says or shows. The other sentences are interpretations that need a stated standard and further evidence. **** **Question 3.** What is triangulation? [( )] Writing a finding in three parts. [( )] Asking three reviewers to read one document. [(X)] Supporting a judgement with at least two kinds of evidence and recording conflicts between them. [( )] Rating an indicator three times. [[?]] Different kinds of source, one judgement. **** Triangulation uses different types of evidence, for example a document and a record sample, so that no single source decides. When sources disagree, the conflict is recorded and tested further. **** **Question 4.** A reviewer reads ten files from a list of 120 and finds three with a gap. What may the reviewer write? [( )] Thirty per cent of all files have the gap. [( )] The issuer has a systemic failure. [(X)] In a sample of ten files chosen by the reviewer from 120, three showed the gap; the sample does not allow a statement about the whole population. [( )] Nothing, because the sample is too small to report. [[?]] State what you did and what it can support. **** A finding states how the sample was chosen, its size and what was seen, and says what it can and cannot support. Small samples still give useful signals and questions for the interview. **** ## Interview technique Interviews test, clarify and enrich the desk findings. They are a form of evidence gathering, not a conversation to build rapport at any cost. ### Purpose of the review interview An interview does three things: - It **tests** whether what the documents describe is how the work is done. - It **clarifies** gaps and conflicts found in the desk review. - It **explores** how people understand their responsibilities. An interview does not decide a finding on its own. What people say is evidence of what they say. It becomes stronger when it agrees with documents or records. ### Prepare the interview Before the interview: 1. Draft a **question plan** from the open questions in the desk review, grouped by criterion. 2. Decide **who** you need to speak to: the person who owns the process, a person who carries it out and, where relevant, a person affected by it. 3. Agree **roles**: who leads, who takes notes, who watches time. 4. Send the issuer the **agenda** and the list of roles you wish to meet, not the full question list, unless your engagement terms say otherwise. 5. Confirm **ground rules**: purpose, how notes are used, confidentiality and whether the session is recorded. Recording needs consent and a basis in the engagement terms. ### Question types - **Open questions** invite description: "Walk me through what happens when a learner appeals a result." - **Probing questions** go deeper: "Can you give a recent example?" "What happened next?" - **Closed questions** confirm a fact: "Is the register kept by the programme office?" - **Evidence requests** ask to see: "Could you show me where that is recorded?" Avoid leading questions ("You do second-mark every file, don't you?"), double questions (two questions in one) and questions that contain your own judgement. ### Listening and neutral behaviour A Lead Reviewer shows interest without signalling approval or disapproval. Say "thank you, that helps me understand" and not "that is good practice". Do not coach: do not tell the issuer what you think they should have done. Reflect back what you heard to check it ("So the form is signed by the programme lead before it is filed. Is that right?"). ### Handle difficult moments - **Vague answers:** ask for an example or a record. - **Defensive answers:** restate the purpose and the fairness of the process; lower the pace. - **Answers that conflict with documents:** do not accuse. Say: "The policy describes X. I would like to understand how that works in practice." - **Over-talkative or silent participants:** invite quieter people by role, not by name, and manage time politely. ### Record the interview Take notes in a standard form: date, participants by role, topic, what was said, what was shown, and follow-up actions. Mark direct quotations clearly and keep them short. Keep your interpretation in a separate column or section. After the interview, the team compares notes on the same day, while memory is fresh. ### Interviews and AI If a transcription or summary tool is proposed, check the engagement terms and obtain consent before use. Treat any generated transcript or summary as a draft that you check against your own notes. Disclose use in the report. **Worked example: Turning a document gap into an interview question** In the desk review of an invented employer-run training academy (the issuer), Lead Reviewer Chidi finds that the appeals policy describes three stages but the appeals log shows only stage one entries for the last year. His first draft question reads: "Why do you never let appeals go to stage two?" He sees three problems. The question assumes a cause, it is leading, and it is accusing. He rewrites it: - Open: "Could you walk me through what happens after a learner submits an appeal?" - Probe: "Can you describe a recent case, without names, and what happened at each stage?" - Evidence request: "Could you show me where stage two outcomes would be recorded?" In the interview, the programme manager explains that stage two decisions are held in a separate file. Chidi asks to see it. The file exists and shows four stage two decisions, which the desk review missed. The finding moves from "gap in the log" to "the log and the stage two file are kept apart, which made the evidence harder to trace." **Practice activity: Rewrite and run a short interview** Part A: Rewrite these three questions so that they are open, neutral and single. 1. "You have a proper induction for all staff, don't you?" 2. "Why is your complaints process so slow, and who is responsible?" 3. "Do you think your assessors are well trained?" Part B: In pairs, one person plays an invented programme manager at a partner training provider, the other plays the reviewer. The reviewer has five minutes to explore how staff induction works, using at least one open question, one probe and one evidence request. A third person observes and notes any leading questions and any signals of approval or disapproval. Swap roles twice.

Facilitator notes

**Live element (120 minutes).** Run the role-play in triads with role cards. Give the interviewee card some planted behaviours: a vague answer, a defensive reaction and a point that contradicts the document. Debrief using the observer notes. Stress the difference between evidence of what people say and evidence of what happens. **Shadow review link.** Participants use the interview observation sheet from this session in the shadow review. **Watch for:** reviewers who fill silence by suggesting the answer.

**Question 1.** Which question is open and neutral? [( )] You do check every file, don't you? [(X)] Could you walk me through how a file is checked before it is closed? [( )] Why don't you check every file? [( )] Is your checking process good enough? [[?]] It invites a description and holds no judgement. **** The open, neutral question asks for a description. The other options are leading, accusing or invite a judgement by the interviewee. **** **Question 2.** An interviewee describes a process that differs from the policy. What should the reviewer do? [( )] Tell the interviewee the policy is not followed. [( )] Ignore it, because the policy is the official evidence. [(X)] Say that the policy describes something different, ask how it works in practice and request a record. [( )] End the interview and write a negative finding. [[?]] Seek to understand, then test. **** A neutral statement of the difference, plus a request for an example and a record, lets the reviewer collect evidence without accusing. A finding is written later, on the evidence as a whole. **** **Question 3.** What is the status of what an interviewee says? [( )] It is conclusive evidence. [( )] It has no value unless recorded. [(X)] It is evidence of what people say, and it is strengthened when documents or records agree. [( )] It must be accepted if the interviewee is senior. [[?]] Think about triangulation. **** Interview statements are evidence of what the person says. Agreement with documents and records strengthens them. Seniority does not change the evidence weight. **** **Question 4.** What should the issuer receive before an interview, unless the engagement terms say otherwise? [( )] The full list of questions. [( )] A draft of the findings. [(X)] The agenda and the roles the team wishes to meet. [( )] Nothing at all. [[?]] Prepared people, but not rehearsed answers. **** An agenda and a list of roles let the issuer arrange the right people. Sharing every question risks rehearsed answers, and sharing draft findings would pre-empt the process. **** ## Writing findings A finding is a statement the reader can follow from evidence to conclusion. A recommendation is a suggestion that follows from a finding. Both must be fair to the issuer and useful to the decision-making body. ### The chain from evidence to finding Each finding should be traceable through four links: 1. **Criterion and indicator:** what was tested. 2. **Evidence:** what was seen or heard, with references. 3. **Analysis:** what the evidence means against the indicator. 4. **Finding:** the conclusion, with a rating from the agreed scale. If a link is missing, the reader cannot check the finding, and the issuer cannot respond to it fairly. ### Anatomy of a good finding A well-built finding has: - **A neutral heading** that names the topic, not a verdict. - **A statement of what was examined**, including sample details. - **The evidence**, with references. - **The analysis**, in plain words. - **The rating** from the agreed scale. - **Any mitigating or positive evidence**, so that the picture is balanced. ### Language for fair findings Use precise, neutral wording: - Say what you saw: "In eight of ten files examined, the second marker's signature was present." - Avoid emotive words: "failure", "negligent", "alarming". - Avoid vague words: "some", "many", "often", unless you give numbers. - Avoid speculation about motives: write "the log did not record" and not "the issuer hid". - Keep to the criteria: do not rate the issuer on matters outside the review scope. ### Fairness checks before you finish 1. Does every finding rest on evidence that the issuer could recognise as its own? 2. Have I reported positive evidence as carefully as negative evidence? 3. Would I write the same finding for a different issuer with the same evidence? 4. Have I separated what I observed from what I infer? 5. Is each recommendation proportionate to the finding? ### Recommendations A recommendation says what would address the finding. Good recommendations are: - **Linked** to one finding. - **Specific** about the outcome, leaving the method to the issuer. - **Proportionate** to the size of the gap. - **Achievable** in a stated period. Reviewers recommend outcomes and do not design the issuer's solution. Designing the solution would compromise independence, since the reviewer would later judge their own advice. ### Use of AI in drafting AI tools can help with grammar and structure, but they can also add unsupported statements. If you use one, check every sentence against your evidence table. Never give an AI tool confidential issuer material without permission under the engagement terms. State in the report which tool category was used and for what task. **Worked example: Rewriting an unfair finding** Lead Reviewer Sofia reads a draft finding from a colleague about an invented industry association that offers a professional badge: "The association clearly fails to manage complaints. Records are poor and the team seems unconcerned." Sofia applies the fairness checks. The draft uses emotive words, gives no numbers, speculates about attitude and does not reference evidence. She rewrites it: > **Finding: Complaint handling records.** > *Indicator tested (illustrative):* Complaints are logged and answered within a stated time. > *Evidence:* The complaints policy (version 2) states a 15 working day response time. The complaints log for the review period contains eight entries (illustrative). Five show a response within 15 working days, one shows a response after 22 working days, and two show no response date. In the interview, the membership manager said that responses are sent by email and not always entered in the log. > *Analysis:* The policy sets a clear time limit. The log shows that the limit was met in five of eight entries. Two entries cannot be assessed because no response date is recorded. > *Rating:* Partly met (using the scale in the engagement brief). > *Recommendation:* Ensure that every complaint response is recorded with its date so that compliance with the stated time limit can be shown. Every claim now has a reference. The rating is traceable. The recommendation names an outcome. **Practice activity: Write a finding and a recommendation** Using the evidence table you built in the desk review practice activity (Lakeside Skills Association, complaints indicator), write one finding of 120 to 180 words using the structure above. Then write one recommendation of no more than 40 words. Finish by running the five fairness checks and noting your answer to each in one line. Swap with a peer. The peer marks any sentence they cannot trace to evidence.

Facilitator notes

**Live element (90 minutes).** Provide three weak draft findings (emotive, vague, speculative). Participants rewrite them in groups, then compare. Run a short "evidence trail" game: a participant reads a finding and others must say where each claim comes from in the evidence table. Anything without a source is struck out. **Watch for:** recommendations that tell the issuer how to run its business.

**Question 1.** Which element must appear in every finding so that the reader can check it? [( )] The reviewer's personal opinion of the issuer. [( )] A comparison with other issuers by name. [(X)] The evidence with references, and the analysis that links it to the indicator. [( )] A forecast of the decision. [[?]] A finding must be traceable. **** Evidence with references and clear analysis make a finding checkable. Opinions, named comparisons and forecasts have no place in a finding. **** **Question 2.** Which sentence suits a fair finding? [( )] The association is negligent in handling complaints. [( )] Many complaints seem to go unanswered. [(X)] In the log for the review period, two of eight entries have no response date recorded. [( )] The team obviously does not care about complaints. [[?]] Prefer numbers and observable facts. **** The correct sentence states a number and an observable fact. The others use emotive, vague or speculative wording. **** **Question 3.** A reviewer wants to write: "The issuer should buy a case management system and train two staff to run it." Why is this a problem? [( )] Recommendations must never be specific. [( )] The recommendation is too short. [(X)] It designs the issuer's solution, which weakens independence, since the reviewer may later judge their own advice. [( )] It does not mention the decision-making body. [[?]] Recommend the outcome, not the method. **** Reviewers recommend outcomes and leave the method to the issuer. Prescribing a solution also steers the issuer towards specific products, which a vendor-neutral review should avoid. **** **Question 4.** What is the purpose of reporting positive evidence in a finding? [( )] To soften a negative rating. [( )] To meet a word count. [(X)] To give a balanced and accurate picture so that the rating is fair. [( )] To please the issuer. [[?]] Accuracy cuts both ways. **** Fair findings reflect all relevant evidence, positive and negative. The aim is accuracy, not softening or pleasing. **** **Question 5.** Which AI-related practice is acceptable in drafting a report? [( )] Pasting the full issuer file into a public tool to speed up drafting. [( )] Accepting generated sentences without checking them. [( )] Leaving AI use out of the report to avoid questions. [(X)] Using an approved tool for grammar checking, verifying every statement against the evidence table and disclosing the use in the report. [[?]] Permission, verification and disclosure. **** AI may assist with approved tasks if the engagement terms allow it. The reviewer checks every statement against evidence and discloses the use. Confidential material is not pasted into unapproved tools. **** ## Ethics Independence and confidentiality are the basis of trust in the Quality Network. This module builds on ICB-R04 and applies it to your daily review practice. ### Independence in practice Independence means that your judgements are based only on evidence and criteria, and that a reasonable observer would see no reason to doubt it. It has two parts: **independence in fact** (you are free of influence) and **independence in appearance** (others have no reasonable grounds to doubt it). Typical threats are: - **Financial or personal interest** in the issuer or a competitor. - **Prior involvement**, such as advising the issuer on the area under review. - **Familiarity**, from a long relationship with the issuer's staff. - **Pressure**, from the issuer, from colleagues or from time constraints. - **Gifts and hospitality** that create a sense of obligation. Your declaration covers these threats. If a threat arises during the review, you inform the coordinator at once and follow their instruction. When in doubt, declare. ### Confidentiality in practice Reviewers see material that the issuer has not made public: policies, data, names and internal discussions. Basic rules: - Use material **only** for the review. - Store it **only** where the engagement terms allow. - Share it **only** with team members and the coordinator. - Do not discuss the review with outsiders, including in social media or casual conversation. - Remove or mask **personal data** in working notes where possible. - **Return or delete** material at the end, as the terms require. Confidentiality has limits. If you see evidence of serious harm, illegality or risk to learners, do not wait for the report. Raise it with the coordinator at once, as the engagement terms and applicable law require. ### Use and disclosure of AI AI tools raise specific ethical points in review work: - **Confidentiality:** issuer material must not be sent to tools that are not approved for the engagement. - **Accuracy:** AI output can contain errors and invented details. You check it. - **Accountability:** you remain responsible for every judgement and sentence in the report. - **Bias:** a tool may reproduce biases in its training material. Do not use it to rate issuers. - **Disclosure:** the report states which tasks used AI, which tool category and how output was checked. Do not use AI to generate ratings or findings. Use it, if at all, for low-risk support tasks that the engagement terms allow. ### Fairness to the issuer Fairness is an ethical duty as well as a quality requirement. It includes giving the issuer a chance to respond to the facts, treating all issuers by the same method, and reporting positive evidence as carefully as negative. It also includes courtesy: reviewers are guests in the issuer's organisation. ### Handling pressure An issuer representative may suggest that a negative finding could damage the organisation. A colleague may suggest a softer rating to keep the relationship friendly. Respond calmly. State that the report follows the evidence and the criteria, and that the issuer can use the factual accuracy step to correct errors. Do not negotiate ratings. Record the approach and inform the coordinator. **Worked example: An invitation and an unexpected disclosure** Lead Reviewer Marta is on site at an invented employer-led training academy (the issuer). At the end of the first day, the academy director invites the review team to a dinner at a restaurant, and mentions that the academy would welcome Marta's advice on its new quality manual "informally, outside the review". Marta considers both threats. The dinner is hospitality that could create a sense of obligation. The advice request is prior involvement, since she would later judge the manual she helped to shape. She replies politely: "Thank you for the invitation. We have to decline hospitality during a review, and we cannot advise on matters within the review scope. Our coordinator can answer questions about the process." She records the exchange and informs the coordinator the same evening. The next day, a staff member mentions in passing that learner files are kept on a shared drive open to the whole organisation. Marta notes the remark, says she will raise it with the coordinator, and does not look at the drive. The coordinator advises on the route for the concern. Marta records it in her log and does not discuss it with anyone outside the team. **Practice activity: Three scenarios** For each scenario, write a short response (60 to 100 words) that says what you would do, what you would say and whom you would inform. 1. You discover that a cousin of yours is the finance manager at the issuer you are about to review. 2. The issuer's quality lead emails you a draft of the factual accuracy response and asks whether you could "adjust the rating slightly" if they add one more document. 3. A fellow reviewer proposes to paste the issuer's full complaints log, including learner names, into a free public chat tool to get a quick summary. Compare answers in your cohort and identify any step that all responses share.

Facilitator notes

**Live element (90 minutes).** Run the three scenarios as case discussions in groups of four, then debrief in plenary. Ask groups to rank the scenarios by risk and defend their ranking. Link back to ICB-R04 materials. Invite participants to share, without names, a real situation where they felt pressure. **Watch for:** answers that handle the issue privately without informing the coordinator.

**Question 1.** During a review, an issuer offers the team a hospitality package. What should the Lead Reviewer do? [( )] Accept if the value is small. [( )] Accept, and mention it in the report. [(X)] Decline politely, and inform the coordinator and record the offer. [( )] Accept if the issuer is a long-standing contact. [[?]] Appearances of obligation matter. **** Hospitality can create a sense of obligation. Declining and recording protects independence in appearance. The coordinator is told so that the matter is handled consistently. **** **Question 2.** An issuer asks for informal advice on a document that falls within the review scope. What is the best response? [( )] Give brief advice and note it in the log. [( )] Give detailed advice, since it will improve quality. [(X)] Decline to advise on matters within the review scope and direct the issuer to the coordinator for questions about the process. [( )] Give advice only after the interviews. [[?]] Remember who will later judge that document. **** Advising on matters you will judge creates self-review and compromises independence. Process questions go to the coordinator. **** **Question 3.** Which statement about confidentiality is correct? [( )] Review material may be shared with colleagues who are interested in the topic. [( )] Confidentiality ends when the report is signed. [(X)] Review material is used only for the review, shared only with the team and coordinator, and returned or deleted as the terms require. [( )] Confidentiality applies only to personal data. [[?]] Think of purpose, access and end of the engagement. **** Confidential material is used only for its purpose, with access limited to those who need it, and handled at the end as the engagement terms say. It covers more than personal data. **** **Question 4.** A reviewer sees evidence of a serious risk to learners during a review. What is the right action? [( )] Keep the matter confidential until the report is issued. [( )] Tell the issuer's director privately. [(X)] Raise it with the coordinator at once, as the engagement terms and applicable law require. [( )] Post a warning on a public forum. [[?]] Confidentiality has limits, and there is a route for concerns. **** Serious risks are escalated promptly through the route in the engagement terms. Waiting for the report, or raising it outside the route, could harm learners or breach confidentiality. **** ## Assessment and badge ### Assessed task The assessed task has two parts. A trained assessor scores both against the reviewer rubric below. This file does not grade your work. **Part 1: Full review of a case file with report.** You receive a case file for an invented issuer (provided by your facilitator). The file contains an application, policies, records and an interview schedule with interview notes or a recorded role-play. You: 1. Write a desk review plan. 2. Complete an evidence table that applies the illustrative indicators supplied with the case. 3. Prepare an interview question plan and, where the cohort format provides one, conduct the interview. 4. Write a report with findings, ratings and recommendations. 5. Include your declaration of independence and confidentiality and your AI use statement (even if the statement says no AI was used). **Part 2: One shadow review.** You shadow a Lead Reviewer on a real or simulated review, as arranged by your facilitator. You observe the planning, desk review, interview and reporting stages and complete a shadow review record that notes: - what you observed at each stage, - two decisions the Lead Reviewer made and your reasoning on them, - one thing you would do the same and one thing you would do differently, - how independence and confidentiality were handled. Facilitators arrange shadow reviews so that the shadowed reviewer is not from your own organisation, to protect independence. ### Reviewer rubric Each criterion is scored by the assessor on four levels: Not yet, Developing, Secure, Strong. | Criterion | Linked outcome | Not yet | Developing | Secure | Strong | |---|---|---|---|---|---| | Application of criteria and indicators | 1 | Criteria are missing or misread; ratings have no stated basis | Criteria are used but ratings are inconsistent between similar items | Indicators are applied consistently with a stated basis for each rating | Application is consistent, calibration notes show reasoning, and bias threats are addressed | | Desk review quality | 1, 2 | Evidence is described without references or quality checks | Evidence is referenced; quality checks are partial; sampling is not explained | Evidence table is complete, with references, quality checks and stated sampling | Evidence is triangulated, conflicts are recorded, and open questions are carried to the interview | | Interview technique | 2 | Questions are leading or closed; the plan does not follow from the desk review | Questions are mostly open; some leading or double questions remain | Plan follows from the desk review; questions are open, neutral and single; evidence is requested | Technique adapts to answers; difficult moments are handled neutrally; records separate statements from interpretation | | Findings and recommendations | 3 | Findings lack evidence or use emotive language; recommendations prescribe solutions | Findings are mostly evidence-based; some vague wording or untraceable claims | Each finding is traceable from indicator to evidence to rating; recommendations are linked and proportionate | Findings are balanced, precise and checkable; positive evidence is reported; recommendations state outcomes and timeframes | | Independence and confidentiality | 4 | Declarations are missing or handling of material is unsafe | Declarations are present but incomplete; handling is mostly safe | Declarations are complete; material is handled as the terms require; concerns are escalated correctly | Threats are anticipated and managed; pressure scenarios are handled calmly; the AI statement is complete and accurate | | Shadow review record | 2, 4 | Record lists events without reflection | Record shows observations and some reasoning | Record explains two decisions, compares own practice and notes how ethics were handled | Record gives insightful, evidence-based reflection and links observations to the criteria | ### Pass rule To pass, you must: - reach **Secure** or higher on every criterion in the rubric, and - complete both parts of the assessed task, and - achieve **at least 8 of 10 correct answers** in the final knowledge check below (illustrative threshold). If you do not pass a criterion, the assessor gives written feedback and one chance to resubmit that part. ### Evidence submitted for the badge - The **case report** (Part 1), including the desk review plan, evidence table, interview plan, findings and recommendations, and your declarations. - The **shadow review record** (Part 2). The badge "Global Quality Seal: Lead Reviewer Training (Ability, L4)", alignment code ICSF-QUAL-L4-A, records that you met the assessment criteria for this course. It does not prove competence in any job. ### Final knowledge check **Question 1.** Which order describes how a reviewer should reach a judgement? [( )] Evidence, criterion, indicator, judgement. [(X)] Criterion, indicator, evidence, judgement. [( )] Judgement, criterion, evidence, indicator. [( )] Indicator, evidence, judgement, criterion. [[?]] Begin with what good looks like. **** Starting from the criterion and indicator keeps the judgement tied to a stated standard. Starting from the evidence invites confirmation bias. (Outcome 1.) **** **Question 2.** Two reviewers disagree on a rating of the same evidence. What is the best response? [( )] The more senior reviewer decides without discussion. [( )] Record the average rating. [(X)] Discuss against the indicator wording, agree a shared reading and write a calibration note. [( )] Ask the issuer to choose. [[?]] The aim is a repeatable reading. **** A calibration note records the shared reading so that later judgements are consistent. (Outcome 1.) **** **Question 3.** What can a policy alone show in a desk review? [( )] That the rule is applied in every case. [(X)] That a rule exists; records are needed to show whether it is applied. [( )] That the issuer meets the indicator in full. [( )] That the issuer's staff understand the rule. [[?]] Intent versus practice. **** A policy shows intent. Records and interviews test application. (Outcomes 1 and 2.) **** **Question 4.** A reviewer reads ten files from a list of 120 and finds a gap in three. Which statement is acceptable in a report? [( )] Thirty per cent of all files have the gap. [(X)] In a sample of ten files chosen by the reviewer from 120, three showed the gap; the sample does not allow a statement about the whole population. [( )] The issuer has a serious systemic problem. [( )] No finding can be made. [[?]] State the method and the limits. **** A good report states how the sample was chosen, what it showed and what it cannot support. (Outcome 2.) **** **Question 5.** Which question is best suited to a review interview? [( )] You do second-mark every final assessment, don't you? [( )] Why is your assessment process so unreliable? [(X)] Could you walk me through how a final assessment is checked before the result is released? [( )] Do you think your assessors are good enough? [[?]] Open, neutral, single. **** The open, neutral question invites a description and carries no judgement. (Outcome 2.) **** **Question 6.** An interviewee describes a practice that differs from the written policy. What should the reviewer do first? [( )] Tell the interviewee that the policy is not followed. [( )] Record a negative finding straight away. [(X)] Note that the policy describes something different, ask how it works in practice and request a record. [( )] End the interview. [[?]] Understand, then test. **** Neutral exploration plus a request for a record produces evidence without accusation. A finding is written later on all the evidence. (Outcome 2.) **** **Question 7.** Which finding wording is evidence-based and fair? [( )] The association is negligent in handling complaints. [( )] Many complaints seem to go unanswered. [(X)] In the log for the review period, two of eight entries (illustrative figures) have no response date recorded. [( )] The team obviously does not care about complaints. [[?]] Numbers and observable facts. **** Fair findings state observable facts with numbers, and avoid emotive, vague or speculative language. (Outcome 3.) **** **Question 8.** Why should a reviewer avoid designing the issuer's solution in a recommendation? [( )] Because recommendations must always be vague. [( )] Because the issuer will not read specific advice. [(X)] Because the reviewer might later judge their own advice, which weakens independence. [( )] Because the decision-making body does not read recommendations. [[?]] Think of self-review. **** Recommending the outcome and leaving the method to the issuer protects independence and fairness. (Outcomes 3 and 4.) **** **Question 9.** During a review, an issuer invites the team to dinner. What should the Lead Reviewer do? [( )] Accept, since the dinner is outside working hours. [( )] Accept only if the cost is small. [(X)] Decline politely, record the offer and inform the coordinator. [( )] Accept and declare it in the report. [[?]] Appearance matters as much as fact. **** Hospitality can create a sense of obligation or the appearance of it. Declining and informing the coordinator protects independence. (Outcome 4.) **** **Question 10.** Which approach to AI in review work is acceptable? [( )] Paste the issuer's complaints log into a public tool to summarise it quickly. [( )] Accept generated findings if they sound reasonable. [( )] Omit AI use from the report if the output was edited. [(X)] Use only tools and tasks allowed by the engagement terms, check all output against evidence and disclose the use in the report. [[?]] Permission, verification and disclosure. **** You remain accountable for every judgement. Confidential material stays out of unapproved tools, output is verified and use is disclosed. (Outcomes 3 and 4.) **** ## Sources and further reading Status as of November 2026; check before publishing. The Global Quality Seal and the ICoBC VRITE criteria are **[Emerging]**. Indicators are not yet published, and internal documents use different expansions of the acronym. Always use the current versions supplied in your engagement brief. **ICoBC materials** - ICoBC Credentialing Skills Framework (ICSF), skill QUAL "Quality assurance and accreditation". Available from ICoBC, info@icobc.net. - ICoBC engagement brief, criteria document and pilot guidance for your review. Issued by the review coordinator. Record the version of each. - ICoBC course ICB-F09 Quality in Credentialing: How Trust Is Built. - ICoBC course ICB-R04 Independence and Conflicts of Interest in the Quality Network. **Standards bodies and frameworks** - International Organization for Standardization (ISO). ISO 19011, Guidelines for auditing management systems. Check the current edition with ISO or your national standards body. - International Organization for Standardization (ISO) and International Electrotechnical Commission (IEC). ISO/IEC 17021-1, Conformity assessment: Requirements for bodies providing audit and certification of management systems. Check the current edition. - European Association for Quality Assurance in Higher Education (ENQA). Standards and Guidelines for Quality Assurance in the European Higher Education Area (ESG). Check the current version on the ENQA website. These sources describe audit and review practice in neighbouring fields. They offer useful background on evidence, independence and reporting, but they are not ICoBC requirements. Where they differ from your engagement brief, follow the brief.