Capstone: an evidence-based strategy and defence

Defend a useful strategy, including its limits

The capstone integrates the course into a decision another person can understand, challenge and use. Produce a strategy for your authorised organisation or the fictional community advice service. Both routes have equal standing. A rigorous test design using fictional evidence is a complete route; a live test is optional and requires separate organisational authorisation, appropriate professional review and resources. Do not spend, send, publish or collect live data merely to make the submission look more authentic.

Your task is to decide what the organisation should do over 90 days to advance its mission through better marketing choices. Show how the diagnosis, audience, offer, support, communication, AI use, measurement and delivery plan fit together. You are not required to use every channel or adopt AI. You are required to explain consequential choices and what would make you change them.

The complete brief

The fictional advice service can accommodate 40 attended appointments each month. Its tracked monthly cohort has 600 unique service-page visitors, 72 enquiries, 24 bookings and 15 attendances. The improvement project has a maximum £1,000 and 20 staff hours over 90 days. A colleague proposes more social posting and predicts a 50% attendance increase. That prediction is unsupported. The course's other fictional notes and scenarios can inform your analysis, but do not combine unrelated datasets as if they were one observed study.

Choose a provisional explanation for the access problem, compare serious alternatives and recommend a feasible next course of action. Include consequences for people who use the service and those not reached by the current route. Distinguish appropriate non-use from avoidable exclusion. Do not count attendance as proof of useful advice or claim that unused capacity proves demand.

For an authorised real-organisation route, replace the case baseline only with appropriately permitted, non-sensitive information. State provenance, missing data and permission boundaries. If essential information is unavailable, use a design with explicit assumptions or return to the fictional case rather than inventing results. A proposed role is not an accepted appointment, a budget is not spending authority and a drafted plan is not a delivered project.

Build one coherent submission

The core strategy is 2,500 words, with supporting tables and appendices. State how you counted words. Include an executive recommendation, problem and evidence, audience and behavioural diagnosis, strategic alternatives, service/offer, support and communication choices, responsible AI/manual workflow, measurement, budget and delivery. Appendices may hold source logs, calculations, instruments, checked outputs and the detailed 90-day plan, but do not hide the decisive argument there.

Carry forward three earlier assessment artefacts: the 1,500-word evidence-and-diagnosis piece, the 1,500-word critical comparison and the 1,000-word AI workflow evaluation. Revise them for consistency rather than rewriting them solely to generate volume. If later evidence changes an earlier conclusion, explain the change and preserve the trail. Cross-reference repeated evidence instead of claiming that the same fictional observation independently confirms several hypotheses.

Your final strategy and defence make up 40% of the self-study portfolio. Evidence and diagnosis contribute 20%, critical comparison 20% and AI workflow evaluation 20%. These weights structure self-review; they do not constitute external marking or an accredited award. No independent academic level has been established. If a future host offers human assessment, its availability, funding, criteria and status must be stated separately.

Stage 1 — reconcile evidence and scope, 8 hours

Spend two hours inventorying the portfolio, two checking sources and calculations, two revisiting the diagnosis and two identifying boundaries and missing prerequisites. Produce a short evidence map that connects each major claim to its source and limitation. Preserve the distinction between authored case facts, inferred mechanisms, external research and untested proposals.

Select three uncertainties that could alter your recommendation. Rank them by consequence rather than convenience. State which can be investigated within the 90-day project and which require a narrower claim. Check the 20-hour allowance against all proposed improvement work; the 40 capstone study hours are learner time, not extra staff hours for the organisation.

Stage 2 — choose and resource the strategy, 10 hours

Allocate three hours to serious alternatives, three to audience/offer and channel integration, two to budget and two to ownership and recovery. Produce a choice memo with at least two rejected options described fairly. Show conditions under which a rejected option would become better.

Build a cash and person-hour budget with uncertainty and a reserve. State routine operating assumptions separately. Assign proposed roles to decisions, review, maintenance and safe stopping. Include an offline or assisted access consideration and a no-paid-tool/no-AI route. If a proposed feature or provider is unavailable, identify a viable manual alternative rather than making the course depend on a subscription.

Stage 3 — design learning and challenge the plan, 10 hours

Use four hours for the test protocol and measurement plan, three for critique and three for revision. A test protocol needs a hypothesis, mechanism, primary measure, denominator, time window, comparison, permission boundary, capacity and balancing checks, and continue/revise/stop rules. State the limits of causal interpretation. You may simulate decisions with the supplied data, but label every simulated result.

Critique can be self-led. Write the strongest objection from a service user, an operational owner and a sceptical evidence reviewer. Then answer each using evidence or a revision, not confidence. If a consenting peer is available, they can ask the questions, but no community membership or live meeting is required. Do not claim this replaces independent expert review of the course.

Stage 4 — write, defend and hand over, 12 hours

Allocate six hours to the final strategy, two to checking the whole pack, two to the defence and two to the final revision/handoff. Check consistency across baseline, resource totals, claims, owners and timing. A promise that appears only in the communication appendix can still invalidate the strategy if it is unsupported.

Give a 10-minute defence, live or recorded if you choose, or complete the written question-and-answer alternative. No upload of a recording is required. For a solo oral route, read the questions aloud, record locally if useful and review your answers against the rubric. Essential evidence must be available in text. The written alternative answers the same six questions in approximately 800–1,000 words; the format is an access choice, not a lower standard.

Six defence questions

1. Which observation most influenced your diagnosis, and what does it not establish? 2. What is the strongest competing explanation, and why did you not choose its intervention first? 3. Whose experience is missing or could be made worse by the strategy? 4. How do cash, staff hours and operational capacity constrain the recommendation? 5. What did your AI/manual evaluation establish, and what remains untested? 6. What result, incident or missing prerequisite would make you revise or stop, and who acts next?

In a 10-minute defence, use roughly one minute for the recommendation and the remaining time for these questions. Do not treat memorised fluency as evidence of judgement. A good answer may acknowledge that an important point remains unknown and explain the safe consequence. After the defence, write a short revision note showing what changed, or why no change was warranted after considering a specific objection.

Completion and next steps

Self-completion means you have read the required lessons, completed the exercises and portfolio, applied the rubrics, corrected material errors and completed a defence or written alternative. It does not mean the organisation's plan has been implemented, outcomes achieved, legal compliance established or expertise independently certified. Keep a local record of the route used and the limits of the evidence.

This authored course is versioned 1 November 2026. Source access, licences, academic suitability, practitioner relevance and accessibility of a future delivery environment need independent review. Report suspected errors through the course host's available contact route; if none is provided, retain a correction note with the relevant lesson and claim.

If you want to explore how Charitas might support your organisation, an optional discovery conversation is a separate next step through its public contact route where available. It is not required for lessons, resources, completion or continued access. You do not need to become a Charitas customer, subscribe to software or share your portfolio. No personal tutoring or funded assessment is promised by this invitation.

Submission checklist

Prepare: (1) revised 1,500-word diagnosis with research plan and evidence log; (2) revised 1,500-word critical comparison; (3) reproducible AI/manual comparison with checked artefacts and 1,000-word evaluation; (4) 2,500-word strategy, budget, measurement protocol and 90-day plan; (5) 10-minute defence or approximately 800–1,000-word written answers to the six questions; (6) revision/handoff note. Supporting tables can be plain text. Use local files or paper; no platform upload is required by this content pack.

Staged work and decision gates

At the end of Stage 1, check claim provenance and select a provisional diagnosis. At Stage 2, reject options that exceed capacity or need unconfirmed promises. At Stage 3, challenge causal and ethical assumptions, then revise. At Stage 4, recalculate totals and explain the plan without relying on appendices for its central argument. These are self-review checkpoints, not new human permission gates for learning. Real external action still requires the organisation's relevant authorisation.

Manual and access routes

Use supplied fictional cases, original internal reading comparisons and simulated AI artefacts. A notebook/calculator suffices. If an external source is unavailable, state that limitation and do not imply you read it. If speaking or recording is unsuitable, use the written defence. If peer feedback is unavailable, write and answer the strongest objections yourself. No paid software, advertisement, model access, customer signup or live pilot is required.

Apply the same five criteria to each assessment component: evidence and source appraisal 25%; critical reasoning 25%; contextual fit 20%; feasibility and evaluation 20%; clarity and reflective judgement 10%. For each criterion score 0–4, then calculate sum(weight × score / 4). Calculate the portfolio learning score as diagnosis × 0.20 + comparison × 0.20 + AI evaluation × 0.20 + strategy/defence × 0.40. Within the final component, judge the strategy and defence together: answers must support or revise the written choices, not earn points for performance style. These scores are self-review aids, not independently validated grades or an award threshold.

Evidence: claims trace to sources, fictional facts remain labelled, and limitations are material rather than ceremonial. Reasoning: compare serious alternatives and answer a strong objection. Fit: connect audience, service, support and communications to the mission and affected people. Feasibility: reconcile £1,000 and 20 hours for the fictional project, with operational capacity, recovery, stop rules and measurement. Clarity/reflection: a reader can understand the decision and see why it changed.

At 0, work is absent or materially misleading. At 1, it asserts a plan without adequate evidence. At 2, it offers partial support but leaves consequential gaps. At 3, it makes a coherent conditional choice with a feasible plan and explicit limitations. At 4, it tests its strongest rival explanation, handles contrary evidence, accounts for excluded perspectives and specifies a credible reversal/handoff. A critical error cannot be averaged away: correct fabricated research, unauthorised personal information, unsupported public promises and infeasible budgets before self-completion. There is no invented pass mark or claim of expert certification.

Annotated strategy extract

'Investigate eligibility and enquiry handling before expanding reach. The fictional cohort has 72 enquiries and 24 bookings, but we do not know how many non-bookings were appropriate resolutions. Reserve six project hours for evidence work, four for an access prototype, three for review, three for test planning, two for handoff and two for contingency. Keep the £1,000 cash ceiling, including an uncommitted £300 reserve. Do not promise new appointment times before operational confirmation. A later reach intervention becomes preferable if current access is usable and eligible demand remains low.'

This is stronger than 'post more to fill all 40 places' because it identifies a decision, alternatives, capacity and reversal condition. It is not a full strategy: it still needs the source appraisal, service/communication choices, responsible workflow and test protocol. Do not copy it as proof of completed research.

Annotated defence answer

Question: 'What evidence would change your mind?' Weak: 'I am confident this is best.' No reversal condition. Competent: 'If eligibility is understood, I would consider promotion.' A useful condition but an incomplete response. Strong: 'If an appropriately bounded audit and enquiry design show that most non-bookings were suitable resolutions rather than access failures, access repair loses priority. I would compare partner-led discovery with broader promotion, checking handling capacity before any launch. That evidence has not been collected in this fictional submission.' The answer preserves status and shows how the decision would change.

Illustrative self-review arithmetic

A hypothetical final-component profile of 3, 3, 4, 3, 4 yields 25×3/4 + 25×3/4 + 20×4/4 + 20×3/4 + 10×4/4 = 82.5/100. This illustrates calculation, not a mark awarded to the extract. If component scores were 70, 75, 65 and 82.5, the weighted portfolio score would be 75/100. No academic pass, credit or level follows from that figure.