Rubrics
What each deliverable is marked against, and by whom. Read the sheet before you start. It is the specification, not a summary of one.
How to read these
Every sheet has the same shape. Learn one and you can read all seven.
Each instrument is split into two halves. Colby marks every design criterion on every deliverable; Ryan marks every business criterion. Ownership follows the criterion, not the assignment. Both instructors mark every submission, and no deliverable in this course is only a design exercise or only a business one.
Weights inside a sheet are shares of that instrument, not of the band and not of the course. A band marked 85% has rows totalling 85.
The gates at the bottom of each sheet are conditions, not criteria. They are not scored and they are not marked down. A submission missing one comes back unmarked.
Three instruments are scored on criteria: the Design System, the Business Validation Report, and the Showcase and Process Book. Those are the three where a mark has to hold up if you challenge it. The other four arebanded completion checks: a tick list, then one judgment per section in four bands. That is not pass/fail. A band of 2 is not a zero. It is the appropriate way to mark work whose real question is whether it was done, and done properly.
Value Proposition Canvas
A completion check, not a criterion rubric. The question is whether the canvas was done and done properly, so each half is banded once rather than scored across six criteria. The canvas is the first thing you hand in and it is the spine of everything after it: W03 asks you for the riskiest assumption on it, and W04 goes and tests that.
A banded completion check: a tick list and one judgment per section, not a criterion score.
Take these with you
A completion check, not a criterion rubric: two sections, each banded once. The tick list is the floor. The band is the judgment.
- Customer jobs stated as jobs, not as demographics
- Pains are specific enough to design against
- Gains distinguished from the absence of pains
- The canvas is legible. A reader can follow it without you
- Fit is claimed explicitly, not left implied
- One customer segment, named and narrow
- Pain relievers map onto stated pains, one to one
- Gain creators map onto stated gains
- At least one element traceable to a person actually spoken to
- Assumptions marked as assumptions
- The riskiest assumption is identified
Required to pass, not scored
Conditions, not criteria.
- AI Use Statement attached
- Submitted at the start of W03, not during it
Market Research & Business Model Canvas
A completion check, not a criterion rubric. Two artifacts marked as one: the market research and the Business Model Canvas built on it. The inherited brief asks for eight parts and a five-source minimum, and both are reflected below. The due date is proposed rather than agreed. It is not in the 2026 schedule table and returns from 2025.
A banded completion check: a tick list and one judgment per section, not a criterion score.
Take these with you
A completion check. Two artifacts marked as one: the research, and the canvas built on it.
- The BMC is readable as one page, not nine disconnected boxes
- Findings are presented, not pasted
- Where a chart is used, it is the right chart
- Five or more sources, cited with URLs
- Market size stated with the method that produced it, not just the number
- Competitor set named, with what each one actually does
- All nine BMC blocks populated
- Every BMC block marked as evidenced or assumed
- Pricing has some basis beyond preference
- At least one finding that contradicts what you expected
Required to pass, not scored
Conditions, not criteria.
- AI Use Statement attached
- Sources listed with working links
Project Proposal
A completion check, not a criterion rubric. This is the Problem/Opportunity Validation Report under its 2026 name, and the inherited brief carries an evidence floor: five or more interviews, three or more market-size data points, two or more current-solution costs, one or more LTV estimate. That floor is the tick list. The band is whether the evidence behind it is real.
A banded completion check: a tick list and one judgment per section, not a criterion score.
Take these with you
A completion check. The inherited evidence floor is the tick list. The band is whether the evidence behind it is real.
- The problem is stated as a specific deficiency, not a general aspiration
- A reader who was not in the room could act on it
- The document is structured and navigable
- What is proposed is scoped to the weeks actually remaining
- Five or more interviews, with evidence they happened
- Three or more market-size data points
- Two or more current-solution costs
- One or more lifetime-value estimate
- Jobs to be done articulated
- Existing workarounds documented
- Willingness to pay addressed with something other than a guess
Required to pass, not scored
Conditions, not criteria.
- AI Use Statement attached
- Evidence appendix included
Design System & Prototyped Solution
The largest single deliverable in the course. You are handing in two things that are really one thing: a small design system, and the prototype visibly built from it. The system is not documentation. It is the set of decisions that made the prototype quick to build and consistent to look at. The floor is deliberately low and deliberately hard: colour with named roles, two type families, one spacing unit, three components, and three to five rules written as sentences. Nothing here rewards volume. What is being marked is whether the decisions are yours, whether they came from your evidence, and whether the thing you built actually obeys them.
Scored on criteria, four levels each.
Take these with you
The brief says what to do. The marking sheet says what it is assessed on. The self-assessment is the same sheet for you to fill in first. Read the brief before you start, not after.
Conceptual Development
20%- The visual direction is argued from the venture’s evidence, not from taste
- Alternatives were generated and rejected, and the rejected ones are shown
- The direction travelled: the final is demonstrably not the first output
- States a non-goal: an aesthetic deliberately not chased, and who you are content to lose
Visualization & Production
35%- Colour has named roles (ink, ground, accent, states), not a mood board
- Contrast is measured and the figures are recorded, not asserted
- Two type families at most, on a scale of four or five sizes with the jumps stated
- One spacing unit, and the multiples actually in use
- Three components, shown in their real states, chosen because they carry this prototype
- The prototype is visibly built from the system, with no orphan screens
Communication
15%- Three to five rules, written as sentences someone else could apply
- Each rule is falsifiable: you could point at a screen that breaks it
- A stranger could rebuild one screen from the system without asking you a question
Evidence Handling
15%- Names where directed generation worked, and where it produced something confident and off-brief
- Shows what was checked, and against what source
- The correction is visible in the work itself, not only described
Research & Validation
8%- The system’s decisions trace back to customer evidence from W02–W05
- Claims about what this audience needs to feel are grounded in someone actually spoken to
Project Management
7%- The scope named at W07 is the scope delivered at W08
- The cut list is real: what was dropped is named, not quietly missing
Required to pass, not scored
Conditions, not criteria. A submission missing one of these comes back unmarked rather than marked down.
- AI Use Statement attached
- The W07 visual context submitted alongside the system
- Contrast checked, with the figures recorded
- Prototype reachable by both markers: a link or file that opens
Business Validation Report
Five pages, one per section, each carrying a fifth of the business half of this sheet. The criteria below are the Fall 2025 rubric v1.1, preserved: the same twenty-five criteria, same equal weighting, in the same order. What each one describes is what full marks looks like; the four levels behind it are the same sentence with less of it delivered. The one addition is a design half, because a validation report has a communication craft to it and the inherited sheet never assessed that.
Scored on criteria, four levels each.
Take these with you
The Fall 2025 rubric v1.1, preserved: the same twenty-five criteria at the same equal weighting, with a design half added for the report’s communication craft.
Communication
8%- The report is readable by someone who has not followed this venture: structure, signposting, and a summary that stands alone
- Each section follows from the one before it rather than restating it
- Claims are legible as claims: what is known, what is estimated, what is assumed
Visualization & Production
7%- Tables, charts and financial exhibits are readable and correctly labelled
- The document is built from your own design system, like everything else you made
- Data is shown where a number would be harder to read than a picture, and not otherwise
Business Model Validation Status
3.4%- All 5 elements (Customer Segment, Value Proposition, Customer Relationships, Revenue Streams, Key Resources) addressed with accurate classification and supporting evidence
Evidence Quality
3.4%- Evidence is specific, cited, and directly supports validation claims
Validation Insights
3.4%- Insights show genuine learning: concept changes, surprising behaviors, invalidated assumptions with pivots
Remaining Hypotheses
3.4%- Hypotheses clearly articulated with testing plans, timelines, and resource needs
Evidence Handling
3.4%- Which model assumptions the agent supplied, and which ones you actually tested, with the validated, partially validated and hypothetical kept apart
- Shows what was checked, and against what source
- The correction is visible in the work itself, not only described
Target Customer Analysis
3.4%- Primary segment defined with market size estimate and validation evidence
Customer Acquisition Strategy
3.4%- CAC estimates for 1+ channels with rationale
Value Proposition & Positioning
3.4%- Core value proposition with customer quote; competitive positioning and differentiators articulated
Marketing & Sales Strategy
3.4%- Pre-launch, launch, retention, and referral strategies specified
Launch Plan
3.4%- Month 1, 3, 6 goals with measurable success metrics
Roadmap (6–18 months)
3.4%- 3 phases with team size and key hires (or justification if none)
Operational Requirements
3.4%- Technology, supply chain, QC, and customer support addressed
Resource Requirements
3.4%- Funding by phase, partnerships, and equipment/facilities identified
Risk Assessment
3.4%- 3 risks (scaling, resource, market) identified with specific mitigations
Feasibility & Coherence
3.4%- Plan is realistic given stated resources and timeline
Unit Economics
3.4%- CAC, LTV, LTV:CAC ratio, and payback period with clear assumptions
Revenue Model Validation
3.4%- Pricing strategy with evidence; break-even analysis
Revenue Projections
3.4%- 3-tier projections (conservative/realistic/optimistic) with rationale
Cost Structure Analysis
3.4%- Variable costs, fixed costs, gross margin, contribution margin specified
Cash Flow Projections
3.4%- Burn rate, months to break-even, startup cash, funding statement complete
Investment Thesis
3.4%- 5 elements: investability, market size, competitive advantage, execution, return potential
Critical Success Factors
3.4%- 4 factors: what must go right, key assumptions, essential resources, market conditions
Next 90 Days Action Plan
3.4%- Week 1-2, Month 1, Month 2, Month 3 with measurable actions
Long-term Vision (2–3 years)
3.4%- Market position, revenue target, team size, exit strategy addressed
Strategic Coherence
3.4%- Recommendations follow logically from validation evidence and financial model
Required to pass, not scored
Conditions, not criteria. The appendix is assessed inside the criteria above rather than separately weighted, but its absence is a condition.
- AI Use Statement attached
- Sources section with citations and URLs
- Interview evidence or links included
- Supporting graphs and charts present where a claim depends on them
Showcase Presentation
Five minutes, solo, in front of an industry panel, on the last day of the term. Eighteen of these run back to back, so the format is enforced rather than policed. At five minutes you will be stopped, finished or not. What is being marked is not polish. It is whether a room of people who have never met your venture can tell, in five minutes, what problem it solves, who for, what you learned that changed it, and why anyone should believe you.
Scored on criteria, four levels each.
Take these with you
Marked live, in the room, while you present. One page on purpose. A sheet a marker cannot fill in five minutes is a sheet that does not get filled.
Communication
20%- The argument is legible to someone who has never seen this venture
- Opens with the problem, not with the solution or the technology
- Built for five minutes, not a fifteen-minute talk delivered quickly
- Nothing on a slide that is then read aloud
Visualization & Production
20%- The deck and the artifact are visibly built from your own design system
- Legible from the back of the room. This is a measurable claim, not a taste one
- What is shown is the work itself, not a description of the work
Evidence Handling
10%- Can say, if asked, which part of this was model-assisted and what you checked before putting it on a slide
- Shows what was checked, and against what source
- The correction is visible in the work itself, not only described
Research & Validation
20%- Claims about customers are traceable to people actually spoken to
- Names at least one thing that testing changed
- Distinguishes what is validated from what is still a hypothesis
Viability
20%- There is a stated model for how this makes money, at a scale that matches the claim
- The ask is specific: what this venture needs next, and from whom
- Survives the panel’s questions without the answer changing
Project Management
10%- What was promised across the term is what arrived
- The cut list is visible in what was presented. Scope was managed, not abandoned
Required to pass, not scored
Conditions, not criteria. The running order depends on all of them, which is why they are firmer here than anywhere else in the course.
- Five minutes. The format is enforced, not policed. You will be stopped.
- AI Use Statement attached
- Slides submitted the night before, for the running order and the tech check
- Present for the whole session, not only your own slot
Process Book
The long form of the showcase: the whole term’s process, carrying the research, the iterations and the decisions behind what you presented. It follows the process-book convention you already know from the Design Cores, so the format needs no teaching. One thing does: a process book defaults to chronology, and chronology is the weakest possible version of it. What makes this worth reading, and worth marking, is the record of what you chose and what you rejected, with the reasoning. Organise it that way and the section headings do most of the work.
Scored on criteria, four levels each.
Take these with you
Marked afterwards, from the document. Organise the book by decision rather than by week and most of this sheet answers itself.
Conceptual Development
20%- Rejected alternatives are shown, not just the path taken
- The reasoning behind each turn is recorded, not reconstructed afterwards
- The venture visibly travelled: the December version is not the September one
- Includes at least one thing you were wrong about, and how you found out
Communication
20%- Organised by decision, not by week
- A reader who was not in the room can follow the whole arc
- The book is designed. It obeys your own system, like everything else you made
Evidence Handling
10%- The design work the model shaped is named, term-wide, and so is what you rejected from it
- Shows what was checked, and against what source
- The correction is visible in the work itself, not only described
Research & Validation
25%- The evidence arc is complete: discovery, testing, and what each one changed
- Sources are cited and interview material is present, not summarised away
- Shows the assumptions that did not survive
Project Management
15%- The dated cut list from W08 is here, with what happened to it
- Scope decisions are recorded as decisions, with dates
Evidence Handling
10%- Which market and customer claims the model supplied, and which of those you actually tested
- Shows what was checked, and against what source
- The correction is visible in the work itself, not only described
Required to pass, not scored
Conditions, not criteria.
- AI Use Statement attached
- Submitted as one document, in a format both markers can open
- Every source cited. The appendix is part of the book, not an optional extra
Participation
There has never been a participation instrument in this course. This is one. It is not attendance. Attendance is a condition below. What is banded is what you brought to the room: the standup that opens every session, and whether your presence changed anyone else's work. Published in W01, which is what makes it markable.
A banded completion check: a tick list and one judgment per section, not a criterion score.
Take these with you
Published in week one, which is what makes it markable. Banded from the standup log at the end of term, not from memory.
- Standup contributions are specific: the number that moved, not a status update
- Gives usable critique in studio, not only praise
- Work is visible in progress, not only when finished
- Peer micro-teach delivered when it was your table's turn
- Brings evidence to the room rather than opinion
- Reports what testing actually said, including when it was unwelcome
- Engages with other ventures, not only their own
- Present and useful at the two live events, UAT in W09 and showcase in W12
Required to pass, not scored
Conditions, not criteria.
- Attendance at W09 UAT and W12 showcase. Both are live and cannot be made up
- A standup log entry for each session attended
The full scheme
Seven instruments. Where a rubric is not published yet, it has not been written. This table says which.
| Instrument | Course | Design | Business | Rubric |
|---|---|---|---|---|
| Value Proposition Canvas | 10% | 30% | 70% | Banded checklist |
| Market Research & Business Model Canvas | 10% | 20% | 80% | Banded checklist |
| Project Proposal | 10% | 40% | 60% | Banded checklist |
| Design System & Prototyped Solution | 20% | 85% | 15% | Scored rubric |
| Business Validation Report | 20% | 15% | 85% | Scored rubric |
| Showcase Presentation | 12% | 50% | 50% | Scored rubric |
| Process Book | 8% | 50% | 50% | Scored rubric |
| Participation | 10% | 50% | 50% | Banded checklist |
Across the whole course that works out at 44% design and 56% business. That balance is a decision, and the generator checks it on every build.
These sheets are generated from a single specification, which also draws the PDF. The criteria above and the criteria on the downloaded sheet are the same text, produced by the same build.