Week 13 · From Insight to Campaign Recommendation

Two weeks from submission, this studio is where your team finds its gaps while they are still cheap to fix. Score yourselves against the rubric, premortem your recommendation, exchange feedback with another team, and rehearse your pitch. The final artefact of the module is your Defensible Recommendation.

At a glance


Time	The full Unit 12 studio session — 85 minutes across four rotations (25 + 25 + 30 + 5)
Work with	Your capstone team, with peer and facilitator feedback
You need	Your own capstone artefacts — orchestration log, dashboard, story, and draft recommendation — plus the Session 12 rubric self-assessment and pitch feedback form (Session folder), and another team to trade feedback with
You hand in	A studio checkpoint: completed rubric self-assessment, peer feedback, timed pitch rehearsal, and a specific action list — filed with your project folder
Graded on	Nothing here is marked directly. It is a honest self-check: score evidence, not effort — for every claimed strength, ask where a marker would find it

The scenario

No dataset today. The inputs are your own capstone artefacts.

Part 1 — Rubric self-assessment · 25 min

Before feedback rounds begin, score your project against the four marked dimensions (25% each) — honestly, and against evidence, not effort.

Dimension	Excellent	Good	Developing	Weak
1. Data handling, agent orchestration & supervision	Clean, traceable pipeline. The log shows real catches and corrections of agent mistakes at multiple stages, each verified. Governance and data-residency handled thoughtfully.	Sound pipeline with a mostly complete log. At least one genuine correction. Verification present but uneven.	Pipeline works but the log is thin or written after the fact. Little evidence of catching agent errors.	Agent output accepted largely unchecked. No meaningful log or verification.
2. Visualisation & dashboard design	Honest, audience-fit charts. The dashboard answers the stakeholder’s question directly, with no distortion or clutter.	Mostly clear charts, right types, minor design or honesty slips.	Some poor chart choices, or a dashboard that shows data without answering the question.	Misleading or confusing visuals, or a chart dump.
3. Narrative & audience fit	One clear message, structured for the named stakeholder, owned by the team rather than the agent. The pitch lands.	Clear story with a defined audience. Some structure or delivery weaknesses.	Findings presented without a clear through-line or audience awareness.	No narrative. A list of outputs.
4. Recommendation & business impact	A specific, defensible recommendation the stakeholder could act on, sized where possible, with limits acknowledged.	A sensible recommendation, lightly evidenced or lightly caveated.	A vague or generic recommendation not clearly tied to the analysis.	No real recommendation, or one the data does not support.

For each dimension, ask the guiding question — Dimension 1: does the log show you directed and checked the agent at every stage? Dimension 2: are your charts honest and the layout clean? Dimension 3: is the pitch owned by you or the agent? Dimension 4: can the stakeholder actually act on your recommendation? — then note one thing you will improve before final week for each.

Premortem · 25 min

Complete the Session 12 premortem worksheet (Session folder). Assume it is three months after your recommendation was implemented and it failed. Write the failure story first, then work backwards to what you can still check, caveat, or change before submission.

Part 2 — Peer feedback · 25 min

Swap projects with another team. You have ten minutes to walk through their work and give one praise, one question, and one suggestion for each dimension:

Dimension 1 — Data handling and orchestration
  Praise:     [something specific in their log or dashboard that shows supervision]
  Question:   [what would help you understand their supervision choices?]
  Suggestion: [one way to strengthen the log or verification work]

Dimension 2 — Visualisation and dashboard design
  Praise:     [one chart or dashboard feature that is clear and honest]
  Question:   [what does this chart make you curious about?]
  Suggestion: [a chart choice, layout change, or distortion they could fix]

Dimension 3 — Narrative and audience fit
  Praise:     [a moment where the story landed or the audience connection was clear]
  Question:   [who is this for, and is that clear from what you just saw?]
  Suggestion: [how the pitch could be tighter or more compelling]

Dimension 4 — Recommendation and business impact
  Praise:     [is the recommendation specific and actionable?]
  Question:   [what would the stakeholder do with this? is that clear?]
  Suggestion: [should it be bigger, smaller, more caveated, or more specific?]

Then swap roles and return the favour.

Part 3 — Pitch rehearsal · 30 min

Rehearse against another team, timed.

Section	Speaker(s)	Time allocated
Opening hook		1 minute
Finding 1 + chart		1.5–2 minutes
Finding 2 + chart		1.5–2 minutes
Finding 3 + chart		1.5–2 minutes
Recommendation and call to action		1 minute
Total		5–7 minutes

Before you start:

Dashboard is open and every chart visible
Each speaker has their section written or outlined
A timer is running
Someone is noting rough times at each transition

Deliver the pitch end to end without stopping — you are testing timing and flow, not perfection. The other team completes the Session 12 pitch feedback form, scoring clarity, pace and energy, stakeholder fit, evidence, and whether the recommendation is specific and actionable. Their hardest question becomes your homework.

Revision plan · 5 min

Before you leave, record the specific action list:

What we will fix before final week (two or three things):


When we will rehearse again:

Who will lead the revision:

Verify & submit

Every self-assessment score points at evidence a marker can open
The three lowest-scoring subcriteria have named owners and a fix date
The premortem produced at least one caveat that is now IN the recommendation
Someone outside your team has asked your recommendation its hardest question, and you have an answer or a to-do
The recommendation says what you decided, not what the agent produced

None of this is a formal submission — file the rubric self-assessment, peer feedback sheet, and revision plan with your project folder. The marked artefacts (dashboard, orchestration log, portfolio, pitch, peer evaluation) are due later, on the published assessment schedule. The full marking rubric and the orchestration log template are in the Capstone pack (Assessment Pack folder in Google Drive).

Final-week readiness checklist

Complete this the week before you present — every team member signs off that the team is ready.

Dashboard

Live in Data Studio (or your chosen tool); every filter and chart works without errors
Titles and labels are clear and grammatically correct
No chart is included unless it answers the stakeholder’s question
Colours and layout pass the five-second test
Tested on another device or browser

Orchestration log

All six pipeline stages documented (Frame, Wrangle, Explore, Visualise, Model, Communicate)
Each stage records what you asked, what the agent produced, what you accepted or changed, and how you verified
At least three verified corrections appear across more than one stage
A comparative hosted-vs-self-hosted governance decision, safeguards, decision boundary, and AI disclosure are complete
The log is on a single sheet or clearly organised across pages

Individual Analytics Portfolio

Each member has drafted their portfolio (2,000–3,000 words of their own writing, plus linked artefacts)
Follows the Individual Analytics Portfolio Brief and covers all five evidence sections
Written in the author’s own voice, not copied from team notes
Spell-checked and proofread

Peer evaluation

Each member has completed the peer evaluation, submitted through xSiTe

Pitch

Rehearsed end to end at least twice; total time 5–7 minutes (timed)
Each speaker knows their section without reading word-for-word
Opening hook is one compelling sentence
Recommendation is specific and actionable; all three findings have evidence
Addresses the named stakeholder directly

Materials for the panel

Dashboard link or PDF ready to share
Portfolio and peer evaluation complete in xSiTe; PDF exports ready if needed
Drive Project Hub and orchestration-log links accessible to the panel
Each member has a copy of the rubric

Governance and disclosure

Where the data was processed and which platforms were used is written down
If you used a China-hosted backend (such as MimoClaw), this is stated with an explanation of why it was acceptable for your data
A plain sentence discloses that AI agents were used and that the team verified the output — on the dashboard, in the pitch, or in the orchestration log

Companion links: Unit 12 app · lesson plan