WF-25 — Alumni Outcomes and Licensure Tracking

This single workflow carries 30 percentage points, more than any other in the pilot, and it is the least formalised process in the entire architecture. It also has the longest lead time of anything here, because graduate outcomes cannot be reconstructed years later. A graduate who left in 2024, whose university email has lapsed, whose personal contact details were never captured and whose employer was never recorded, is not a data-quality problem that can be fixed with an extract. That cohort is gone. Every month this workflow does not exist is a month of outcomes that will never be recoverable, and the three-year rolling averages on all three primary KPIs mean the loss persists for three further cycles after it happens.

Two other things about this workflow are unlike anything else in the pack. First, all three primary KPIs are Track B: the Ministry computes them, and IoL influences them through contact data, response rates and, crucially, through the guide's explicit invitation to submit its own data where that data is more accurate. That invitation is the strongest indirect tactic available anywhere in this pilot and section 9 develops it in full. Second, this workflow handles personal data about people who have left the institution, in larger volume and over longer periods than anything else here, which makes data protection an operating constraint rather than a compliance footnote.

1. Purpose

To ensure that IoL knows, for every graduate of PGDipHPE and MScHPE, what happened to them: whether they are employed or in further study at twelve months, in what role and for which employer, whether that employer is an organisation where they trained during their studies, and whether they have obtained a professional licence by three years after graduation; and to hold that knowledge as dated, per-person, evidenced records that can be produced on demand, so that the Ministry's employment and outcome KPIs rest on what IoL actually knows about its graduates rather than on whoever happened to answer a survey.

2. Scope

Scope statement. This process manages graduate outcome data from the capture of contact and consent at graduation clearance, through the twelve-month outcomes collection, the employer identification and the three-year licensure sweep, to the supply of contact data to the Ministry, the optional submission of substitute data, and the assembly of the KPI 3.1 numerator and denominator.

Applies to. Every graduate of PGDipHPE and MScHPE, from the point at which they complete graduation clearance, for a minimum of three academic years afterwards and, for the alumni relationship generally, for as long as consent holds. It applies to the graduate outcome data itself, to the contact data supplied to MoHESR for the Graduate Destination Survey and the Employer Satisfaction Survey, to the employers-of-graduates list, and to professional licence attainment.

Does not apply to. The placement episode itself, which is WF-13; this workflow receives the placement organisation identity from it and returns the twelve-month outcome to it. The design and administration of the employer and supervisor feedback instruments, which is WF-14; this workflow supplies the employer population that WF-14 surveys for KPI 2.4. The micro-credential half of KPI 2.5 and the classification of a programme as tied or not tied to a mandatory licence, which are WF-22; this workflow collects the licence attainment data that the licence half of KPI 2.5 needs and does not decide whether that half applies. Graduation clearance itself, which is WF-10; this workflow attaches its capture step to it. Alumni engagement, fundraising and community-building activity, which are outside the pilot.

Applicable requirements. OBEF guide v11.5 KPI 1.1, KPI 1.2, KPI 3.1 and the licence component of KPI 2.5, with their Appendix A anchors; Appendix C sampling and small-population rules, noting that Appendix C is truncated in the source PDF at "1 of 4"; Appendix B evidence provisions, which name contact data for graduates confirming professional licences and raw data and metadata for any substitute data submitted in place of centrally administered surveys, covering KPIs 1.1, 1.2, 2.4, 2.6 and 3.1; CHEDS Students - Graduates.xlsx, Graduate Licensure.xlsx, Institute - Employers.xlsx and Institute - Academic Programs.xlsx; MBRU data protection policy and UAE data protection law as they apply to former students [IoL to confirm the applicable instrument and the lawful basis relied on]; MBRU records retention schedule; MBRU alumni relations policy [IoL to confirm].

The definitions that govern everything below.

KPI 1.1, employment rate. The percentage of graduates in full-time employment or further study within twelve months of graduation, from the cohort that graduated twelve months before reporting time, on a three-year rolling average. Full-time means at least 32 hours per week, or the professional norm for the sector. Employment includes formal contracts, salaried posts and documented self-employment, in any sector. Further education means enrolment in an accredited programme or a professional qualification. Anchors: 0, 35, 65, 90, 100 per cent.

The clause that matters most for MBRU, and its limits for IoL. For professions requiring mandatory post-graduation training, which includes health professions, the twelve-month employment window is defined as twelve months after the end of any mandatory internship or training period imposed by a UAE regulator as a licensure condition. Without this, a health-professions graduate serving a mandatory internship at the twelve-month point can read as unemployed. The clause must be asserted with evidence, not assumed, and the evidence is a per-programme licensure-pathway record showing which regulator imposes which training period and for how long.

And here the honest reading for IoL specifically, which differs from MBRU's undergraduate clinical programmes. IoL's learners are practising health professionals, mostly already employed and mostly already licensed, who study health professions education as a postgraduate qualification. PGDipHPE and MScHPE do not lead into a mandatory post-graduation training period, and IoL's graduates are not serving one. The clause therefore protects MBRU's medicine, dentistry and other clinical programmes far more than it protects IoL's. Asserting it for an IoL programme where no graduate is in mandatory training would be asserting it without evidence, which the charter places outside the boundary. Where an individual IoL graduate is genuinely in a regulator-imposed training period, the clause applies to that individual and the evidence exists. Say so, and say only that.

KPI 1.2, employment in relevant jobs. The same population, filtered to full-time employment relevant to the field of study. Anchors: 0, 30, 60, 90, 100. The denominators of KPI 1.1 and KPI 1.2 must be identical, which is cross-KPI assertion C1. Relevance is determined by the Ministry, not by IoL, against a matching matrix built from ENSCO and ISCO occupational classifications and ISCED fields of study, and the ISCED mapping derives from the programme's CIP coding. IoL's only lever on this KPI is that the CIP code is accurate and current.

KPI 3.1, job offer post work-placement. The percentage of graduates who received a confirmed job offer, or were hired, by an organisation where they trained during their studies, from the cohort that graduated twelve months before reporting time. The denominator is graduates from the same cohort who completed at least one placement, restricted to GDS respondents where GDS data is used and extended to all such graduates where IoL submits its own data. Anchors: 0, 10, 30, 50, 100, so the start-of-high anchor is unusually low and reaching 50 per cent already scores 75.

The licence component of KPI 2.5. Eligible graduates from the cohort that graduated three academic years before the reporting period who had obtained a professional licence by reporting time, over all eligible graduates in that same cohort. The cohort is the three-years-prior one, not the current one, which the register names as the most common error on this KPI. Approved mandatory licences relevant to MBRU are the DHA practice licence, the DoH Abu Dhabi licence, the MoHAP licence and the Community Development Authority social care licence.

KPI 2.4, employer feedback in employment. MBRU supplies the employer list; the Ministry distributes the Employer Satisfaction Survey. The list must be the employers of the previous academic year's graduates, not a generic partner list. This workflow owns that population.

3. Trigger, boundary and endpoint

4. SIPOC

5. Accountability

Process owner. IoL Director of Programmes, or the role accountable for the graduate relationship and able to require that outcome capture happens at graduation clearance rather than being attempted afterwards.

R responsible · A accountable · C consulted · I informed. One A per row.

Note the two rows where accountability sits outside IoL's delivery line. The substitute-data decision is accountable to Quality and IQA, because it is exactly the kind of favourable classification the charter requires a second signature for. Personal data retention is accountable to Data Governance, because the volume and sensitivity here exceed anything else in the pilot and the subjects are people who have left and cannot easily object.

Escalation.

6. Process steps

1. Capture contact data, consent and status at graduation clearance, not afterwards. This is the single most important step in the workflow and it costs a form field. [CAPTURE] personal email (not the university address), mobile number, at least one durable alternate contact, current employer and role at graduation, consent to be contacted about outcomes, and consent for contact data to be supplied to MoHESR for the Graduate Destination Survey. Serves 1.1, 1.2, 2.4, 2.5, 3.1. Decision point. Graduation clearance is the last moment at which IoL has the learner's attention and an administrative reason to require a response. A contact detail not captured here will cost between ten and a hundred times as much to recover later, and for some graduates will never be recovered. Make the field mandatory in the clearance checklist (WF-10) and make the personal email distinct from the university one, because university addresses lapse. The consent point, stated properly. The lawful basis for supplying graduate contact data to MoHESR needs to be settled once with Data Governance [IoL to confirm whether this is consent, legal obligation or public task under the applicable instrument]. Whatever the basis, the graduate should be told at clearance, in plain language, that their contact details will be provided to the regulator for the national graduate survey and what that means. Telling them is both the right thing and the thing that makes them answer when the survey arrives.

2. Open the graduate's record in the Alumni Outcomes and Licensure Register. One row per graduate, keyed on Grad_Student_ID, carrying Grad_Academic_Period, Grad_Program_Code, Grad_Major_CIP, Grad_Submission_Term and Grad_Workplacement. [CAPTURE] the register row. Serves all five KPIs.

3. Import the placement history from WF-13. For every graduate, record the organisations at which they trained during their studies, normalised to the controlled organisation list. [CAPTURE] placement organisation identities and the completed-at-least-one-placement flag. Serves 3.1, both numerator and denominator. This is what makes KPI 3.1 computable at all. The KPI asks whether the twelve-month employer is an organisation where the graduate trained. Free-text organisation names on either side make that match impossible, and "Dubai Hospital", "Dubai Hosp." and "DH Dubai Hospital" are three organisations to a matching routine and one to a human.

4. Maintain contactability between cycles. A light annual touch: an alumni update, a CPD offer, an invitation to an IoL event under WF-23. [CAPTURE] last successful contact date, bounce and failure flags, updated details. Serves 1.1, 1.2, 2.4, 3.1. Contactability decays. An email address captured at graduation and never used for three years is a guess. IoL's advantage over a large university is that its programme directors know these people, and that advantage is worth preserving deliberately rather than assuming it will still be there.

5. Determine the Appendix C route and any programme grouping, before any instrument is distributed. [CAPTURE] population size N per KPI per level, the route applied, the required sample or response count, and, where grouping is used, the justification against the three grouping principles, dated and signed. Serves 1.1, 1.2, 3.1 and, through WF-14, 2.4. Decision point, and it is a hard sequencing rule. The grouping decision must be made and documented before the survey runs, never after results are visible. Revising a grouping once the results are known is in the charter's right-hand column. Section 8 sets out the routes and a worked calculation.

6. Supply the graduate contact extract to MoHESR ahead of the GDS launch. [CAPTURE] the extract, its date, its completeness measure and the release log entry. Serves 1.1, 1.2, 3.1. Validation gate. Every graduate row in the reporting cohort must have a non-null, syntactically valid email and mobile, at a completeness of at least 98 per cent. Below that, escalate before the GDS launch rather than after the response rate disappoints. Personal data leaves IoL at this step, so the release is logged, minimised to the fields the Ministry needs, and made under the settled lawful basis.

7. Run the twelve-month outcomes collection. A structured, dated, per-person collection against the OBEF definitions, using the multi-mode attempt protocol in section 10. [CAPTURE] per-person: employment status; hours per week or a professional-norm statement; contract type including documented self-employment; employer name normalised to the controlled list; job title; start date; further-study enrolment and the accrediting body; the date of the enquiry and the date of the response; the mode; and the respondent's own confirmation. Serves 1.1, 1.2, 3.1. Apply the OBEF definitions exactly, and record the raw values rather than the derived flag. Record hours per week, not "full-time", so that the 32-hour test can be re-applied by a second analyst and so that a professional-norm judgement is visible as a judgement. Record the further-study programme and its accreditor, not "yes". Exception route. Where a graduate cannot be reached after the full protocol, record them as non-responding with the attempt log attached. Do not infer their status from LinkedIn, from a colleague, or from what is generally known. An inferred outcome that is later contradicted is worse than an honest non-response, and Appendix B entitles the Ministry to the per-respondent record behind every number.

8. Verify employment where verification is genuinely available, and record what was verified. [CAPTURE] the verification source and date. Serves 1.1, 1.2, and the accuracy claim in section 9. For IoL's population, verification is often available and cheap: many graduates work within Dubai Health, many employers are already known to IoL as placement sites or CPD clients, and a line manager will normally confirm a role. Verification is what turns a self-reported census into a stronger instrument than a self-reported sample, and it is the substance of the accuracy argument in section 9. Verify with the graduate's knowledge, not behind their back.

9. Normalise employer identity and match it to the placement organisations. [CAPTURE] the matched or unmatched determination and the matching basis. Serves 3.1. Decision point. Where the twelve-month employer is an organisation at which the graduate trained, and the offer or hire is confirmed and dated within the twelve-month window, the graduate enters the KPI 3.1 numerator. Where the employer is a different organisation, they do not. Where the graduate was already employed by the placement organisation before the placement, see the open question in section 9 and do not resolve it favourably.

10. Assemble KPI 1.1, KPI 1.2 and KPI 3.1 with a single, shared denominator construction. [CAPTURE] the numerators, the denominators, the denominator basis and the working. Serves 1.1, 1.2, 3.1 and assertion C1. Build the KPI 1.1 and KPI 1.2 denominators once, as one object, and derive both KPIs from it. Assertion C1 requires them to be identical, and the reliable way to satisfy an identity assertion is to make it true by construction rather than to test it afterwards. Record the KPI 3.1 denominator basis explicitly: GDS respondents who completed at least one placement, or all cohort graduates who completed at least one placement. The two give different numbers and the guide requires the basis to be stated.

11. Decide whether to submit substitute data, and route the decision through Quality and IQA. [CAPTURE] the decision, its rationale against the "more accurate" test, and the second signature. Serves 1.1, 1.2, 3.1 and, through WF-14 and WF-06, 2.4 and 2.6. Section 9 sets out the methodology, the documentation pack and the failure modes in full. The decision is made before fieldwork, not after results are known, because a decision made after seeing both numbers is a choice of the more favourable one however it is described.

12. Run the three-year licensure sweep. For the cohort that graduated three academic years before the reporting period, establish professional licence attainment. [CAPTURE] Graduate ID, Licensure/Certificate award status, Licensure/Certificate award date, Licensing body name, Licensing body country, Graduate Program (Program Code). Serves 2.5. Filter to awarded only. Pending and attempted are excluded. For multi-level licences, every intermediate level achieved is counted. The trap, which WF-22 owns and this workflow must not walk into. Most IoL learners already hold a DHA, DoH or MoHAP practice licence, obtained years before enrolment and unconnected to the HPE programme. Those are not KPI 2.5 attainments. This workflow records licence facts, including pre-existing ones and their dates, so that WF-22 can apply the eligibility rule; it does not itself decide what counts. Recording the award date is what makes the distinction possible at all.

13. Supply the employers-of-graduates list to WF-14 for KPI 2.4. [CAPTURE] Name of the employer, Position, Company Name, Email, plus the graduate-to-employer linkage. Serves 2.4. This must be the employers of the previous academic year's graduates, not a partner list. They are different populations and only the first serves KPI 2.4. The register names this as the specific way KPI 2.4 goes wrong at MBRU. Note the programme-level rule that makes the linkage matter: graduates whose employers complete the ESS are linked back to the programmes they graduated from, and those programmes receive the same score the institution received from that employer. The linkage must therefore be recorded and reproducible, not inferred.

14. Retain the raw data, and retire it on schedule. [CAPTURE] the retention class of every record and its disposal date. Serves Appendix B and data protection compliance simultaneously. Appendix B entitles the Ministry to raw data and metadata for any substitute submission, so the raw per-person per-question data with issue and collection dates must be retained. That is a reason to hold it properly, under access control and with a disposal date, rather than a reason to hold it forever.

7. OBEF data generated

Capture rule. Six things cannot be reconstructed afterwards, and five of the six are lost within months rather than years:

• A personal contact route that survives the university email. Captured at clearance or not at all.
• The employer at the twelve-month point, with a date. A graduate who has changed jobs twice by the time anyone asks cannot reliably reconstruct where they were at twelve months.
• The offer or hire date, which the twelve-month rule and any mandatory-training shift both depend on.
• The raw per-person response with its issue and collection dates, which is exactly what Appendix B names for substitute submissions.
• The placement organisation in normalised form, without which the KPI 3.1 match cannot be made.
• The licence award date, which is the only thing distinguishing a licence obtained after graduation from one the learner already held on the day they enrolled.

Reproducibility test. A second analyst should be able to take the Alumni Outcomes and Licensure Register, the controlled organisation list and the documented methodology, and reproduce all three primary KPI values without asking anyone what happened. Today the answer is no, and it is no for the most fundamental possible reason: the register does not exist [IoL to confirm]. There is no capture defect to describe at field level, because there is nothing to describe. This is the one workflow in the pilot where the gap is the whole process rather than an attribute within it, and it is also the workflow with the highest yield. That combination is why the charter puts it first in the build sequence.

8. Max-score design

The thing to understand about KPI 1.1 for IoL, because it changes what the work is. The underlying employment rate is almost certainly at or near 100 per cent, since the learners were employed before they enrolled and remained employed throughout. IoL's risk on a 20-point KPI is therefore entirely a measurement risk, not a performance risk. Every graduate IoL cannot trace, every graduate who does not answer the GDS, and every graduate whose status is recorded ambiguously, is a potential reduction in a number that ought to be perfect. That is an unusual and slightly uncomfortable position: the largest single KPI in the framework will be decided by administrative diligence rather than by anything educational.

Appendix C, the sampling rules, and what they mean for cohorts of IoL's size.

Worked calculation, so the numbers are visible rather than asserted. With Z = 1.645, P = 0.5 and E = 0.10, the unadjusted sample is n0 = (1.645² × 0.5 × 0.5) / 0.10² = 0.6765 / 0.01 ≈ 68. Applying the finite-population correction:

Note the discontinuity, and resolve it in the conservative direction. At N = 70 the formula asks for 35, while the grouping route's stated target for a grouped population is at least 40. A grouped population of exactly 70 therefore attracts two different numbers depending on which rule is read as governing. IoL should use 40, the more demanding, and record the reading. The cost of the extra five responses is trivial and the cost of an under-powered sample is a KPI that may not be usable.

What this means concretely for PGDipHPE and MScHPE. Both cohorts are well below 70 [IoL to confirm the actual annual graduate numbers]. The grouping route therefore applies, and the two programmes plainly satisfy the grouping principles: the same degree level in the postgraduate sense, the same content area of health professions education, and the same employment fields across the health system. Even grouped, the combined cohort may not reach 70, in which case the 50 per cent sample provision applies, or IoL runs a census, which supersedes the question entirely. Note that grouping across graduation years is also permitted by the principles, which may be the route that reaches 70 where combining the two programmes in a single year does not. Whichever route is used, it is decided and signed before fieldwork.

And the honest caveat. Appendix C is truncated in the source PDF at "1 of 4". Further small-sample rules exist and are not available to MBRU. Cohorts of IoL's size are precisely the case those missing pages most probably address. Request the complete appendix from MoHESR in writing, and treat every sampling determination made before it arrives as provisional and dated.

What this workflow must do to reach the target.

1. Build the Alumni Outcomes and Licensure Register. Nothing else in this workflow is possible without it, and it is the number one priority in the institutional register backlog because it serves 32.5 per cent of the framework.
2. Attach contact and consent capture to graduation clearance as a mandatory, non-bypassable step, with a personal email distinct from the university address, a mobile number, and at least one durable alternate contact.
3. Reach 98 per cent contact-data completeness for every reporting cohort before the GDS launch, and escalate below it rather than accepting it.
4. Run a documented near-census at twelve months, rather than relying on whatever the central survey returns from a cohort of this size.
5. Verify employment where verification is available, and record what was verified and how. This is the substance of the accuracy claim in section 9.
6. Normalise every employer name to the controlled organisation list at the point of capture, so the KPI 3.1 match is mechanical rather than interpretive.
7. Construct the 1.1 and 1.2 denominators as one object so assertion C1 holds by construction.
8. Decide and sign the Appendix C route and grouping before fieldwork, every cycle.
9. Sweep licensure for the three-years-prior cohort annually, recording award dates so pre-existing licences can be distinguished.
10. Supply WF-14 with the employers of the previous year's graduates, with named contacts and working emails, and with the graduate-to-programme linkage recorded.
11. Put the KPI 3.1 definitional question to MoHESR in writing, and hold the answer.
12. Treat the personal data properly, because a data protection failure here would end the workflow regardless of its scoring value.

[REDESIGN] actions.

Sequencing note. R2 costs almost nothing and must happen before the next graduation, because every graduation that passes without it is a cohort permanently degraded. R1 is the substantial build and should start in parallel. R3 depends on both and takes a full year to produce its first result, which is why the charter puts this workflow first in the build sequence despite it having no curriculum change to negotiate: it is not slow because it is difficult, it is slow because it is annual.

9. Indirect strategy where data cannot be collected

Categories 1, 2 and 4 all apply here, and this is the section of the whole pack that deserves the most development, because the framework's single most powerful sanctioned provision sits inside it.

Category 1, the Ministry holds the instrument

All three primary KPIs are Track B. MoHESR designs and distributes the Graduate Destination Survey, computes KPI 1.1 and KPI 1.2, applies its own ENSCO to ISCO to ISCED relevance matrix, and computes KPI 3.1 from GDS responses. IoL does not submit these numbers.

The levers are therefore inputs, and there are exactly four.

1. Contact data completeness and currency. A GDS that cannot reach a graduate cannot count them. This is R2 and the 98 per cent gate.
2. Response rate. Graduates answer surveys from institutions they still feel connected to and ignore surveys from institutions they do not. R4's light annual contact is a scoring intervention disguised as an alumni courtesy. Legitimate response improvement, being reminders, timing, framing and a personal approach from a programme director who taught them, is explicitly in the charter's left-hand column. Selecting who is surveyed in order to shape the result is in the right-hand column.
3. CIP coding accuracy, which drives the relevance matrix behind KPI 1.2. This is R9, and it is the only lever on that KPI.
4. Substitute data, which is Category 2 below and is the important one.

What IoL cannot do, and should stop wondering about. IoL cannot see the relevance matrix, cannot appeal an individual relevance determination, and cannot know in advance whether the Ministry will classify a consultant physician's post as relevant to a health professions education qualification. State that as an uncertainty in the forecast rather than modelling a number that looks reasonable. The score plan sets KPI 1.2 at 80 to 95 per cent for exactly this reason and should not be tightened.

Category 2, the population is too small for the instrument, and the substitute-data route

This is the strongest indirect tactic in the whole pilot, and IoL is unusually well placed to use it.

The provision itself. The guide explicitly invites institutions to submit their own data directly to MoHESR for KPIs 1.1, 1.2, 2.4, 2.6, 3.1 and 4.1 where they consider it more accurate than the standard survey. That is an invitation in the framework, not a workaround. Four of those six KPIs are within this workflow's scope or are fed by it, and the six together cover 32.5 percentage points.

Why IoL's case is genuinely strong, and it is worth being precise about why. The argument is not that IoL would like a better number. It is that a central survey of a cohort of IoL's size is a weak instrument and IoL has a better one.

• A national survey sampling a cohort of fifteen to thirty postgraduate learners will typically return a handful of responses. A handful of responses produces a percentage with an enormous confidence interval, which is as likely to understate as overstate, and which cannot meet Appendix C's own thresholds without grouping.
• IoL knows every graduate personally. It is a small, high-contact postgraduate department whose programme directors taught, supervised and assessed each of them, often over two years. That is not a sentimental claim; it is a methodological asset that a national survey contractor does not have.
• IoL can achieve near-census coverage where the GDS achieves a sample, and coverage is the dominant term in the accuracy of a small-population estimate.
• IoL can verify. Many graduates work within Dubai Health, many employers are already known to IoL as placement sites or CPD clients, and employment can frequently be confirmed with the employer rather than resting on self-report alone. A verified census is a stronger instrument than an unverified sample by any standard a reviewer would apply.

The test the guide sets is "more accurate", not "more favourable". Those come apart, and this workflow must be built so that it notices when they do. A census that returns 88 per cent where the GDS sample returned 100 per cent is still the submission to make, because the argument for substituting was methodological and the methodology does not change when the answer is disappointing. A process that substitutes only when substitution helps is not running a methodology; it is running a selection rule, and the selection is the finding a reviewer will make.

The decision rule, therefore, and it is deliberately restrictive. The decision to submit substitute data for a given KPI and cycle is made and signed before fieldwork begins, on methodological grounds, and is not revisited after either result is known. If the pre-committed decision was to submit, IoL submits whatever the census found. If the pre-committed decision was not to submit, IoL does not submit even if the census turns out to look better. The second signature from Quality and IQA is on the pre-commitment, and its date is part of the evidence.

The methodology, which is what makes the submission defensible.

1. Population. All graduates in the reporting cohort, defined identically to the CHEDS Grad_Academic_Period cohort. A census rather than a sample, with the target being complete coverage and the Appendix C route documented as a fallback if coverage falls short.
2. Definitions. The OBEF definitions applied exactly and recorded as raw values rather than derived flags: hours per week against the 32-hour test or the professional norm, with the norm stated where it is relied on; contract type including documented self-employment; further study named with its accrediting body; employer named against the controlled organisation list; job title as the graduate states it, so the Ministry can classify it rather than IoL pre-empting the classification.
3. Timing. Fieldwork anchored to the twelve-month point from graduation, or from the end of a regulator-imposed mandatory training period where one genuinely applies to that graduate, with the anchor recorded per person.
4. Instrument. Aligned to the GDS question set as closely as IoL can establish it, so the two instruments are comparable. Where IoL asks something the GDS does not, that question is marked as additional and does not enter the KPI calculation.
5. Contact protocol. Multi-mode and pre-specified: email, then SMS, then a telephone call, then a personal approach from the programme director, with defined intervals and a maximum. The protocol is applied identically to every graduate, not intensified for those whose answer is expected to be favourable. That symmetry is the single most important integrity property of the whole design and it is also the easiest to lose without noticing.
6. Verification. Where employment can be confirmed with the employer, with the graduate's knowledge, it is confirmed and the confirmation is recorded with its source and date. Where it cannot, the response is recorded as self-reported. Both are legitimate; the distinction is recorded.
7. Non-response treatment. Pre-specified before fieldwork: whether non-respondents are excluded from the denominator, or treated as unknown, or reported separately, and on what basis. This choice moves the number more than almost anything else in the design, which is exactly why it is fixed in advance and recorded. The guide's own KPI 1.1 denominator wording, "survey respondents or otherwise traceable", supports a traceable-population denominator; whichever reading IoL adopts, it adopts it once and holds it across all three years of the rolling average.
8. Calculation. Performed to the Ministry's formula, with the KPI 1.1 and KPI 1.2 denominators built as a single object so assertion C1 holds by construction, and the KPI 3.1 denominator basis stated.
9. Comparison. Where a GDS result also exists, both are reported to MoHESR together with the reasoning, rather than the more favourable one being presented alone.

The documentation pack, which is what survives an Appendix B request. Appendix B names, explicitly, raw data and metadata for any substitute data submitted in place of centrally administered surveys, covering KPIs 1.1, 1.2, 2.4, 2.6 and 3.1. The pack must therefore contain, at minimum:

How this fails, stated so it can be avoided. A substitute submission fails, and takes IoL's credibility with it, if any of the following is true: the methodology was written after the results; the contact protocol was pursued harder for graduates whose answers were expected to help; non-respondents were dropped from the denominator without a pre-specified rule; outcomes were inferred from LinkedIn or from a colleague's recollection rather than obtained from the graduate; the raw per-person data cannot be produced; or the submission was made in the years it helped and withheld in the years it did not. A substitute submission that cannot survive an Appendix B request is worse than no submission at all, because it converts an absence into a finding.

Grouping, the second Category 2 route. Where a census is not achieved and a sample must be relied upon, Appendix C's grouping route applies as set out in section 8. PGDipHPE and MScHPE plainly satisfy the content area and employment-field principles, and grouping across graduation years is available where a single-year combination still falls short. Decided and documented before the survey runs. Never revised after results are visible. That is the whole rule and it does not have exceptions.

Category 4, the activity exists but is invisible

Three distinct invisibilities live in this workflow, and all three are capture problems rather than interpretive ones.

The graduates themselves. IoL's graduates are employed, licensed, senior and, in many cases, still working alongside the people who taught them. Their outcomes are excellent and entirely uncounted, because nobody wrote them down at twelve months. The response is R1, R2 and R3.

The employers. IoL knows who employs its graduates, informally and reliably, and has never converted that knowledge into Institute - Employers.xlsx as an employers-of-graduates list with named contacts and working emails. KPI 2.4 is worth 2.5 points and is currently fed, if at all, by the wrong population. The response is R8.

The licences. IoL's graduates hold and obtain professional licences continuously. Nobody has ever asked the cohort that graduated three years ago what they now hold. The response is R7, and the discipline that makes it usable is recording the award date, so that a licence obtained after graduation is distinguishable from one held on the day of enrolment. WF-22 is explicit that pre-existing clinical licences are not KPI 2.5 attainments, and this workflow's job is to record the facts that let that rule be applied, not to apply it favourably.

Category 3 does not apply, and Category 5 is not yet a supportable reading

Category 3 does not apply: graduate employment is not an activity that fails to exist. Category 5, performance genuinely low, is not a supportable reading for KPI 1.1 or KPI 3.1 before R1 and R3 have run for a cycle, because a low number today would measure the absence of a register rather than the fortunes of IoL's graduates. KPI 1.2 is the one exception: if the Ministry's relevance matrix genuinely does not match a clinical post to a health professions education field of study, that is a real result about classification rather than a capture defect, and the honest response is to report it, understand the matching basis, and consider whether the CIP coding accurately describes the programmes.

The open question, which must not be resolved favourably

Many IoL learners are already employed by the organisation hosting their placement. They are practising professionals undertaking a practicum inside their own health system, frequently inside their own department. KPI 3.1 counts graduates who "received a job offer or were hired by an organisation where they trained during their studies".

Whether continued employment, internal appointment or promotion at that organisation satisfies KPI 3.1 is not addressed anywhere in the guide. WF-13 raises the same question from the placement side. Three readings survive the text:

• Reading A, permissive. Continued employment at the training organisation counts, since the graduate is employed by an organisation where they trained.
• Reading B, restrictive. Only a new confirmed offer or hire counts, since the KPI's evident purpose is to measure placement converting into employment, and an employment relationship that predates the placement was not converted by it.
• Reading C, intermediate. A documented internal appointment or promotion, made after and attributable to the placement, counts; unchanged continued employment does not.

The score plan models this favourably and flags it as an assumption. That flag is the honest position and this workflow must not remove it. Do not assume Reading A.

What to do, concretely. Put the question to MoHESR in writing, in terms that describe IoL's actual population rather than in the abstract, and record the answer against the register and in OBF3.1M. Until the answer arrives, plan against Reading B, the pessimistic case, and record per graduate the facts that let any of the three readings be applied afterwards: whether an employment relationship with the placement organisation existed before the placement began, whether a new confirmed offer, appointment or promotion was made, and its date. Capturing the underlying facts rather than the derived flag is what makes the answer, whenever it comes, applicable retrospectively rather than requiring the cohort to be re-surveyed.

If the answer turns out to be Reading B or C, IoL has an educationally defensible response available independently of the KPI: placing learners, for at least part of the practicum, in an organisation other than their own employer. That is a better educational design in any case, since it exposes a learner to a different teaching culture, and it should be argued on those grounds rather than presented as a scoring manoeuvre.

One further ambiguity worth naming. If IoL's practicum is not mandatory and few graduates complete a registered placement, the KPI 3.1 denominator may be zero or near zero. The guide does not state how a KPI with a zero denominator is treated, and a near-zero denominator produces a percentage so volatile that one graduate moves it by tens of points. This is another reason WF-13's mandatory-practicum redesign matters beyond its own 8 points, and it is a question for WF-26 to route to MoHESR alongside the others.

Boundary check. This workflow must never:

• infer a graduate's employment status from LinkedIn, from a colleague, from a programme director's impression, or from what is generally known, and record it as a response;
• pursue the contact protocol harder for graduates whose answers are expected to be favourable;
• exclude non-respondents from a denominator without a rule specified before fieldwork;
• decide whether to submit substitute data after seeing the result of either instrument;
• submit substitute data in the years it helps and withhold it in the years it does not;
• present a substitute submission it could not evidence with raw per-person data and dates;
• revise an Appendix C grouping after results are visible;
• assert the mandatory post-graduation training clause for a programme or a graduate not actually in a regulator-imposed training period;
• resolve the KPI 3.1 continued-employment question favourably in the absence of a Ministry answer;
• record a placement-organisation employer match on the basis of a name that resembles another name, rather than on the controlled organisation list;
• count a pre-existing clinical practice licence, held before enrolment, as a post-graduation licence attainment;
• use different denominators for KPI 1.1 and KPI 1.2;
• change the denominator basis, the non-response treatment or the grouping between years of a rolling average without recording the change and its effect on the series;
• supply a partnership list to WF-14 in place of an employers-of-graduates list;
• retain graduate personal data beyond the published schedule, share it beyond the minimum necessary, or contact a graduate who has withdrawn consent.

10. Service standards

11. Records and evidence

Appendix B readiness. Today, IoL could produce almost none of this [IoL to confirm]. It could produce the graduate list from Students - Graduates.xlsx and probably university email addresses. It could not produce a twelve-month outcome for any past cohort, a verified employer identity, a licence position for the cohort that graduated three years ago, or raw per-person data with dates. On a workflow carrying 30 percentage points, that is the largest single evidence gap in the pilot, and unlike most gaps in this pack it is not closable retrospectively. The cohorts already lost stay lost. What R1, R2 and R3 protect is every cohort from the next graduation onwards.

Data protection, which is heavier here than anywhere else in this pilot and is not a footnote.

This workflow holds identifiable data about people who have left the institution, over a period of at least three years, covering employment, income-adjacent facts, professional licensure and career progression. Those people have no ongoing administrative relationship with MBRU, are not present to object, and in many cases work inside the same health system as the people holding the data. Six operating rules follow, and they are constraints on the workflow rather than aspirations:

1. Settle the lawful basis once, with Data Governance, for holding graduate outcome data and for supplying contact data to MoHESR, and record it [IoL to confirm the applicable instrument].
2. Tell graduates at clearance, in plain language, what will be held, for how long, who it may be shared with including the regulator, and how to withdraw. A graduate who understands the purpose is also a graduate who responds.
3. Collect the minimum necessary. The KPIs need employment status, hours, employer, job title, further study and licence attainment. They do not need salary, and IoL should not collect it.
4. Apply role-based access. The register is not a departmental shared drive. Access is by role, logged, and reviewed.
5. Distinguish education records from clinical records, since MBRU operates inside a health system and many graduates are clinicians whose licence status also exists in clinical systems governed by different rules. The two must not be conflated or cross-populated.
6. Retire data on schedule. A retention period that is never enforced is not a retention period, and holding graduate data indefinitely because it might be useful is the failure mode this rule exists to prevent.

12. Risks and controls

13. Performance measures

The first three measures decide 30 percentage points and none of them is about employment. They are about whether IoL wrote down a phone number, asked a question at the right time, and kept a record of asking. That is an uncomfortable summary of the largest KPI cluster in the framework, and it is the accurate one.

14. Change control