The Year-Two Mental Model
Everything in Year Two hangs on one structure. Internalize this and every course has an obvious place.
Desirability — prove the opportunity
Two qualitative research waves on Dscout. Who has the problem, how intensely, and why do current workarounds fail? End with an evidence-backed value proposition.
Feasibility — make it tangible
Two research-with-stimuli sprints + a design-review sprint. Choose the bet, put lowest-fidelity stimuli in front of real people, translate into a thin buildable increment.
Viability — prove the economics
Software must be built. Feasibility and willingness-to-pay studies. Expect the "nobody pays $20/month" moment — it's iteration input, not failure.
Convergence — one defensible story
20–25 minute final pitch to 5–7 industry leaders and VCs. Every headline claim carries a proof point, an uncertainty label, a tradeoff, and a decision ask.
The core operating rhythm: Understand → Create → Reflect → Repeat
This loop — not the double diamond — is the capstone's stated operating rhythm. "Create" is defined broadly: prototypes, yes, but also pricing strategies, insight statements, and stimuli. A prototype is not proof of desirability; it's an instrument for learning.
Three frameworks carry the whole year
Desirability / Feasibility / Viability
The grading triad. Start with desirability — "no prize for solving correctly what proves to be the wrong problem" — then layer feasibility (technology landscape) and viability (business context).
Value Proposition Canvas → Business Model Canvas
VPC is the key Q1 deliverable: who you design for, problems solved, why they matter — backed by data. BMC becomes the central viability framework from Q2 through the final pitch.
The evidence ledger (one spine, all year)
One capstone evidence ledger and decision log shared across all nine courses. Each course adds a layer to the same product spine: source, quote/observation, job, implication, confidence, unresolved assumption, next test.
The Operating System
Cross-cutting disciplines every course shares. These are the habits that get graded implicitly all year.
1. Label every claim by evidence status
| Label | Meaning |
|---|---|
| Explicit evidence | Directly present in a supplied source (syllabus, deck, sponsor brief, schedule). |
| Observed finding | What a participant actually said or did — attributed, in context. |
| Interpretation | Your reading of the observation. Not yet an insight. |
| Inference | A useful planning hypothesis (e.g., a course's likely role). Label it or lose credibility. |
| Assumption | Unverified input. Every number in a model is one until sourced or tested. |
| Contradiction | Sources disagree (e.g., MPD 416 Oct 2 AM vs master schedule PM). Preserve, don't silently resolve. |
| Open question | Unknown. Say "not in the provided evidence" rather than inventing. |
2. Ship decision artifacts, not conclusions
Every recommendation carries: assumptions · evidence · tradeoffs · risks · owner · date · kill criteria. A polished conclusion without these is a failed artifact.
3. The checkpoint questions (answer at every gate)
4. Write hypotheses in falsifiable form
Plus: sample, method, numeric or observable threshold, go/no-go/pivot branch, owner, and what changes if the threshold is missed. If the plan can't fail — if no plausible result would change your mind — say so.
5. Evidence discipline — the hard lines
- Never fabricate participants, quotes, counts, market sizes, sponsor constraints, or faculty preferences. Missing evidence stays missing.
- Qualitative claims stay qualitative. Eight interviews are rich evidence, not a population percentage. No prevalence, market size, or significance from a small convenience sample.
- Synthetic interviews and AI-generated reactions are rehearsal, never customer evidence.
- Stated preference ≠ observed commitment. Ask what people did recently under real constraints, not what they'd hypothetically do.
- TAM/SAM/SOM numbers get labeled: sourced, calculated, or "an assumption wearing a decimal point."
- Protect NDA/PII: sponsors own capstone IP; mask participant identity; never put secrets or raw recordings into AI tools.
6. AI guardrails (from the Applied AI course, applied all year)
- Prompts are not controls. Alignment is unsolved; models are "inherently untrustworthy" alone. Deploy AI only inside engineered harnesses with independent guardrails.
- Evals are performance reviews for AI — known-good examples verifying outputs against expectations.
- Model selection is recruiting: match capability to task; diversify providers; treat AI as a variable cost with ROI justification.
- Never place secrets, API keys, or PII in prompts.
- Follow each course's Canvas AI policy — the MPD 416 syllabus permits AI in some areas and prohibits it in others. Disclose and cite as required.
Fall — Prove the Opportunity DESIRABILITY
Fall establishes the opportunity substrate: rigorous research methods (406), the two capstone research waves and opportunity framing (416), and the defensibility of the assets around it (458). Fall pressure is ambiguity and evidence quality, not presentation polish. Exit criteria: one or two evidence-backed opportunities and a clear statement of what remains unproven — otherwise Winter goes solution-first.
Market Research Tools & Methods
Role: methodological engine room under MPD 416's research sprints (planning inference — no syllabus supplied)
Key learning outcomes
- Match the method to the decision. Qual (interviews, ethnography) forms insights and tests IRA; quant (screeners, surveys, sizing) confirms reach, prevalence, and willingness-to-pay once an insight is framed. Don't quant a question that hasn't earned an insight; don't size-justify a market from a small qualitative sample.
- Design research that can fail. State the decision at stake and what positive/negative/mixed evidence would change it. Separate the decision from the research activity.
- Sampling tied to signal. Effect size before n. Power = Signal/Noise; SE = σ/√n with diminishing returns past roughly n = 15–25. State what the least expensive sufficient method is — and what it cannot establish.
- Confidence interval vs prediction interval. CI bounds a population parameter and narrows with data; PI bounds the next individual observation, is always wider, and doesn't shrink with sample size. Conflating them is the Year-1 catapult error relocated to a customer.
- Screeners without bias. Leading language, missing disqualification logic, questions that recruit the answer you want. Recruit for recent/frequent category or JTBD involvement, not generic interest.
- Instrument hygiene. Leading, double-barreled, hypothetical, prestige, recall, and solution-biased questions. Prefer recent behavior, context, workarounds, trade-offs, and neutral "tell me about…" prompts.
- Honest segment sizing. TAM/SAM/SOM logic traced back to evidence quality.
Gut-check questions
- Are eight interviews being treated as qualitative evidence or as a population percentage?
- Who exactly is represented by this sample — and who is missing?
- What is the strongest alternative explanation for this finding?
- Which quote can a teammate trace to the raw source?
- Are TAM/SAM/SOM numbers sourced, calculated, or assumptions wearing decimals?
Customer Driven Opportunities
Role: capstone spine course — Q1 desirability leg
What the course takes you through (syllabus sequence)
| Phase | What you produce |
|---|---|
| 1 · Secondary research (Sep 27–Oct 10) | Log every source: claim, date, method, relevance, bias risk. Synthesize context, existing solutions/workarounds, gaps. Convert unknowns into primary-research objectives. |
| 2 · Qualitative planning (Oct 2–10) | 5–10 question screener with explicit eligibility + one articulation question. Neutral discussion guide with probes, evidence targets, consent, consistent execution. |
| 3 · Wave 1 — problem discovery (Oct 15–23) | Eight 45-minute virtual interviews. Preserve participant-level observations, exact quotes, context, actions, emotions, workarounds, contradictions. |
| 4 · Synthesis (Oct 24) | Observations → codes → themes → insight statements → tensions → scoped HMWs. Findings (what happened) ≠ insights (why it matters). IRA-test everything. |
| 5 · Midterm share-out (Oct 30) | Beachhead persona(s), journey map(s), value-proposition hypothesis — no specific solution. 15-min problem story + 5-min Q&A: WHO/WHAT/WHEN/WHY, functional/emotional/social JTBDs, segments, barriers, evidence. |
| 6 · Opportunity framing (Oct 30 →) | Objective, problem, customer, TAM→SAM→SOM with assumptions, intensity, competition gaps, why now, why us, success measures. Real/Win/Worth to name the weakest leg. |
| 7 · Design-requirements research (Nov 7–25) | Separate outcomes, experience principles, constraints, acceptance criteria, preferences, features. Wave 2 (eight more 45-min interviews) with neutral stimuli: probe behavior, trade-offs, barriers. Record what was confirmed, weakened, or reframed. |
| 8 · Final handoff (Nov 21–Dec 4) | Stand-alone story + Winter handoff: opportunity brief, evidence ledger, requirements, assumptions, gaps, and a falsifiable Discovery Hypothesis. |
Key learning outcomes
- Run real customer discovery: distinguish secondary vs primary, qual vs quant; plan, field, and synthesize two full research waves.
- Synthesis discipline: an insight explains why a pattern exists, connects multiple observations, and reveals a tension between what people want and what exists. An observation is not an insight.
- Workarounds are signals — if people hack the system, the system is mis-designed. The behavioral truth behind a workaround is where opportunities live.
- Beachhead logic: why is this segment the right first audience (intensity, access, learning) rather than merely the largest?
- Value proposition without smuggling in a solution — a customer outcome, stated so it survives solution changes in Q2.
- Judge research quality — a program-level outcome: right questions, right sample, unbiased technique, expected outcomes, required time. You will be evaluated on this even as a non-researcher.
Gut-check questions
- What is the one-sentence WHO/WHAT/WHEN/WHY problem — and does it pass the grandma test?
- Is this an observation, insight, opportunity, requirement, or solution? Have I skipped a step?
- Does the insight pass IRA — and what evidence contradicts it?
- What did Wave 2 confirm, falsify, or force us to reframe?
- Does the story lead with the problem and learning — or is it a solution pitch in research language?
Intellectual Capital Strategy
Role: strategic value & defensibility layer over the opportunity (planning inference)
Key learning outcomes
- IP as system decisions, not legal abstraction. A patent claim-scope decision is a make-vs-buy / asset-specificity decision. A trade-secret-vs-patent choice is disclosure-risk vs enforceability. FTO is risk mapping, like supply-chain exposure mapping. Patent portfolio value is a DCF with dominant terminal-value sensitivity.
- Build an intellectual-capital map: asset → role in customer value → status (owned / controlled / licensed / accessible / shared / exposed) → dependency → copyability → risk → action (build, document, restrict, license, partner, publish, monitor, retire).
- Test the moat mechanism: exclusive access, accumulated learning, trust, data, process know-how, network, switching cost, speed, or legal right — separating evidenced advantage from aspiration. First-to-market ≠ defensible.
- Read a patent landscape like a supply network map: where is the space crowded (commoditized) vs open (differentiated, defensible)?
- Sponsor-IP reality: students sign NDAs; sponsors own resulting IP. What can the team carry forward? What requires counsel review? What can be shown now vs restricted vs not shared?
- Data/AI provenance: permission, privacy, security, retention, model-use constraints, auditability, human oversight.
Gut-check questions
- What is the asset exactly — or is "IP" being used as a vague synonym for a feature?
- Do we own it, control it, have permission, or have temporary access?
- Should we make, buy, ally, publish, keep confidential, or defer — and what objective criteria support it?
- What happens if a partner withdraws, a platform changes terms, or a contributor leaves?
- Are we confusing secrecy with strategy, or technical novelty with willingness to pay?
Winter — Turn Evidence into a Testable Product FEASIBILITY
Winter is the densest quarter: choose and govern the innovation bet (402), generate sprint evidence with real stimuli (403 — the 1-credit studio), and translate the design-review direction into a buildable increment with the McCormick software team (405). The discipline: one product spine across all three so research, prototype, backlog, and metrics accumulate instead of becoming three parallel projects.
Management of Product Innovation
Role: the management layer — choose, sequence, resource, govern, and stop innovation bets (planning inference)
Key learning outcomes
- Turn the 416 output into a small ranked portfolio of bets: 2–4 real alternatives compared on MPD 445's 11-row Decision Criteria Framework — never a single concept scored against itself. Show score assumptions; a rubric is not fake precision's license.
- Protect desirability first. Attractive technology and sponsor preference are not customer evidence. Audit concepts with Gather→Form→Frame and IRA before scoring.
- Real / Win / Worth: name the weakest fit leg and write the falsifiable Discovery Hypothesis with numeric/observable threshold, go/no-go/pivot branch, owner, and date.
- Stop criteria are deliverables. Consensus is not commitment; make uncertainty, ownership, resource implications, and stop/hold/pivot criteria visible.
- Pull every argument into operating reality: tooling lead times, supplier qualification, capacity, regulatory exposure, cost, a plant's willingness to absorb a new SKU. Constraints to investigate, not assumptions to fill in.
- Make/Buy/Ally before building adjacent capability. Run a pre-mortem. Ask which constituency loses budget, headcount, or credit if this succeeds.
Gut-check questions
- What decision are we making, by when, and what changes if we're wrong?
- Is this a real customer job or a feature disguised as a problem statement?
- Which of Real, Win, or Worth is weakest — and what test could change it?
- What 2–4 alternatives did we actually consider, and what objective criteria separate them?
- Where is the unresolved team conflict, and who has authority to close it?
Creativity and Understanding through Design Sprints
Role: the intensive evidence-generating studio — Q2's center of gravity
Key learning outcomes
- Sprint briefs before artifacts: one learning objective, the decision it informs, hypotheses, participant plan, stimulus + fidelity, evidence to collect, risks, stopping rule.
- Lowest-fidelity stimulus that can answer the question: paper, clickable UI, Wizard-of-Oz, landing page, concept video — each with an explicit statement of what it can and cannot test.
- Neutral research: ask for stories and behavior before reactions to the concept. "People liked it" is not evidence of a meaningful job or willingness to act. Enthusiasm, politeness, novelty, and sampling noise are not signals.
- Sprint 2 must deliberately change or contrast the stimulus — never repeat a sprint just to collect favorable quotes. A null result is a result.
- Synthesis with counterevidence preserved: code observations, affinity-map, inspect inconsistencies, trace every insight to attributed source.
- Critique discipline: work, not identity. One specific strength, one sharp checkable ask, owner, next evidence. Process evidence outranks cosmetic polish.
- Design-review pack (Sprint 3): insight → concept → prototype → evidence → open questions; the evidence-to-decision trace; feasibility/viability questions left for Spring; the smallest coherent software handoff.
Gut-check questions
- What exactly can this stimulus teach — and what can its fidelity not teach?
- Did the HMW grow from the insight, or did we write the solution first?
- What did the research contradict? If nothing changed, could the test have changed our mind?
- What is the smallest next test before adding fidelity or features?
- What should 402 decide now, and what must remain open for 405?
Agile Management of Software Product Design & Development
Role: translation layer — design-review learning → buildable increment with the undergrad software team (planning inference)
Key learning outcomes
- The thinnest coherent increment: one user outcome, one learning objective. Explicit in-scope / out-of-scope / deferred / intentionally-manual. A thin slice before broadening scope.
- Full traceability: customer evidence → insight/tension → HMW → concept/prototype → requirement → backlog item → acceptance criterion → observed learning. Every story names the job, evidence, owner, dependency, test, and learning outcome.
- Never blur status: prototype, mock, spike, manual workaround, or implemented capability — say which. Never say "built," "validated," "secure," or "ready" from a plan or screenshot alone.
- MRD vs PRD (MPD 445): the MRD proves the opportunity; the PRD defines execution. User stories as "As a [persona], I want [goal] so that [benefit]" with priority, acceptance criteria, and evidence trace.
- Working agreement with the software team: roles, decision rights, escalation, definition of ready/done, feedback protocol, sponsor touchpoints. Estimates and architecture come from the technical owner — never invented by the PM.
- Learning loop: refinement → pairing → implementation → test → demo → evidence capture → retrospective → decision update. Connect build output to Q3's feasibility and willingness-to-pay tests.
Gut-check questions
- What customer job and evidence does this backlog item trace to — and did design review validate it?
- Is this the thinnest coherent increment, or a feature bundle that feels safer?
- Which requirement is explicit, inferred, or a technical assumption awaiting the team?
- Are estimates and architecture claims from the technical owner, or did we invent them?
- What dependency or decision right is hidden inside this "small" story?
Spring — Prove Viability & Tell One Defensible Story VIABILITY
Spring is the compression point: business-model integration and the final capstone case (409, the 1-credit spine), market story and channels (401), and leadership/alignment under real constraints (449) — all converging on the June 4 pitch. By Spring start, discovery and concept selection should be substantially closed. Spring time goes to viability evidence, economics, alignment, and rehearsal of one decision-ready narrative.
Business Model Design
Role: capstone spine — Q3 viability and final convergence
Key learning outcomes
- A living Business Model Canvas where every block is marked Evidence / Assumption / Test, with the dependency that could invalidate it named. "Which Canvas block is evidence-backed, and which is still a story?"
- Willingness-to-pay as decision evidence: distinguish stated preference, observed commitment, price sensitivity, buyer-vs-user differences, and cost of delivering the whole offer. Define the threshold before the study.
- The $20/month lesson: users rejecting a price is data, not failure. Diagnose — value perception, segment, packaging, pricing unit, adoption friction, or delivery economics — before lowering the number. Then iterate value proposition and pricing model.
- Auditable economics: NPV as the primary investment rule (IRR supporting only). Separate fixed / variable / sunk / opportunity / avoidable costs. Contribution margin, break-even, relevant range, CapEx timing, cannibalization. Sensitivity on the two assumptions that can flip the decision — a model's credibility rests on 3–4 defensible assumptions, not formula complexity.
- Wedge discipline at maximum pressure: one defensible segment, value proposition, and recommended model — not three hedged options.
- The final pitch: every headline claim has a proof point, uncertainty label, tradeoff, and decision ask. Which slide proves the claim, and which merely asserts it?
Gut-check questions
- Is the weakest leg Desirability, Feasibility, or Viability — and what evidence would change that diagnosis?
- What did the WTP study actually measure, and what did it not measure?
- Which two assumptions drive the model, and does the recommendation survive a realistic base/low/high range?
- What whole offer must a mainstream customer receive — and what is Made, Bought, or Allied?
- Where does sponsor/mentor/faculty feedback conflict, and what objective criterion resolves it?
Principles of Marketing
Role: beachhead, positioning, message, channels, market proof (working lenses are planning inferences)
Key learning outcomes
- Beachhead selection as evidence work: compare urgency/intensity, reachable access, WTP, strategic fit, ability to learn. State who is excluded and why. The job defines the segment before demographics do.
- The positioning stack: target customer → problem/job → category/frame → differentiated value → proof. A tagline any competitor could copy is not a position.
- Messages express insights/tensions, not feature lists. What participant quote or observed behavior earns the value proposition?
- Whole Offer / TCE mapping: core product plus onboarding, support, delivery, complementary assets — and the adoption gap that blocks customers beyond early adopters.
- Channels as falsifiable experiments: each route to market gets a hypothesis, method, threshold, cost, owner, leading signal, and kill/pivot criterion. No vanity metrics — every metric needs a decision, baseline, and if-missed action.
- Never market an uneconomic offer: price and message must match WTP evidence, contribution margin, and cost-to-serve.
Gut-check questions
- Which exact beachhead has this problem, in what context — and what evidence shows the intensity?
- What is the customer using today, and what specifically is missing or unacceptable?
- Why now? What changed — and is that claim sourced or merely convenient?
- Which route-to-market can actually reach this beachhead, by whom, at what cost?
- What behavior would falsify this message or channel hypothesis — and what happens if it does?
Global Leadership in a Smart and Connected World
Role: decision rights, stakeholder alignment, global/system risks, leadership evidence (topics are planning inferences)
Key learning outcomes
- Stakeholder & decision maps: interests beneath positions, influence, asymmetric risk exposure, BATNAs, and explicit decision rights for sponsor, mentor, faculty, team, user, buyer, partner.
- Lead without ceding ownership: sponsor and mentor input informs direction; it must not quietly replace the team's evidence-based direction. Tiered options anchored to objective criteria — never arbitrary midpoint compromise.
- Team health as an operating system: Trust → Conflict → Commitment → Accountability → Results. Diagnose which rung is stuck; surface real disagreement (consensus is not commitment).
- Global claims are hypotheses, not stereotypes: define "global" in observable terms; treat any cultural pattern as a context-specific prototype requiring local evidence and individual-level checking.
- Connected-system risk review: access, power, language, privacy, security, AI dependence, data governance, supply/partner dependencies, sustainability — and who carries each risk.
- AI governance: model/agent/harness distinctions, independent guardrails, evals, least-privilege permissions, human approval, logging, failure response, cost/ROI. Prompts are not controls.
- Leadership evidence in the pitch: which leadership claim is demonstrated by an actual decision or trade-off rather than asserted?
Gut-check questions
- What decision is being made, by whom, by when — and what happens if we do nothing?
- What are the other party's interests and BATNA, not the position you're reacting to?
- Which stakeholder carries asymmetric risk, and have you made it visible?
- Which Lencioni rung is the team stuck on?
- Are sponsor/mentor inputs informing direction — or steering the team away from evidence?
Year-1 → Year-2 Framework Crosswalk
Year Two does not introduce a new vocabulary — it runs Year 1's load-bearing frameworks under capstone pressure. Reference them by name; know where each does the work.
| Year-1 source | Framework | Where it works in Year 2 |
|---|---|---|
| MPD 410 · Understanding through Design | Desirability/Feasibility/Viability · Gather→Form→Frame · IRA · HMW scope test · AEIOU/SET · Ten Faces · workarounds as signals | The whole arc's spine. 416 synthesis; 403 sprint framing; every insight IRA-tested; every HMW scope-checked (too narrow / too broad / just right). |
| MPD 400 · Predictive Product Design | Full 8-step JTBD (Define→Locate→Prepare→Confirm→Execute→Monitor→Modify→Conclude) · MoSCoW · Data Radar · RICE with real numbers · Backcasting v0→v1→v2→v3 | 416 personas & JTBD maps; 403 stimulus scoping; 405 backlog and increment planning; 401 job-based segmentation. |
| MPD 445 · Product Management | Real/Win/Worth · 11-row Decision Criteria · Discovery Hypothesis · Whole Offer/TCE · Make/Buy/Ally · MRD/PRD · GTM/AARRR | 402 concept selection; 416 opportunity framing; 405 story discipline; 409 model choice; 401 whole-offer and route-to-market. |
| MPD 432 · Data Analytics | Decision-first · Power = Signal/Noise · C/N/X variables · XmR stability before capability · CI vs PI | 406 sample sizing and instrument design; 403 signal-vs-enthusiasm; 405 metrics; 409 WTP thresholds. No "clearly shows" language the evidence hasn't earned. |
| MPD 424 / 420 · Finance & Accounting | NPV-first · WACC · incremental cash flow · sunk-cost discipline · contribution margin · CVP/break-even · relevant range | 409's economics core; 458's IP valuation; 401's price/margin coherence. Exclude sunk costs; include cannibalization; 3–4 defensible assumptions. |
| MPD 442 · Team Building | Trust→Conflict→Commitment→Accountability→Results · vulnerability loop · belonging cues | 403 critique culture; 405 working agreement; 449 team-health diagnosis — all year, persistent Year-1 team. |
| MPD 444 · Negotiation | Interests vs positions · BATNA vs bottom line · tiered options on objective criteria | Sponsor/mentor touchpoints (449); 402 stakeholder conflict; 458 partner terms; 401 channel partners. |
| MPD 450 · Communication | Minto/BLUF · Bridging · Deficit/Dialogue/Participatory models · Values→Problem→Impact→Reflection→Solutions→Expectations | Midterm and final decks; sponsor readouts; hard conversations when research challenges sponsor assumptions. |
| MPD 408 · Supply Chains | Supply chain surplus · efficiency-vs-responsiveness fit · MOQ as setup-cost response · marginal ROI by class | 458 (asset specificity, exposure); 409 Make/Buy/Ally and operations; 402 operational realism. |
| MPD 430 · Sustainability | Functional unit · system boundary · hotspot analysis · validation for credibility | 409 sustainability claims; 449 connected-system risk review. |
Applied AI Companion
From the capstone's Applied AI opening lecture — the mental models that govern every AI-assisted workflow in Year Two, and echo directly in the capstone's own guardrail language.
The three-layer stack
Model — the "brain," predicts tokens, cannot act. Agent — model + tools + minimal looping toward a goal. Harness — memory, permissions, guardrails, oversight, sub-agents; enables multi-day tasks. Never deploy a bare model.
Managing agents = people management
Nondeterministic systems need crisp goals, success metrics, milestones, and "performance reviews" (evals). Model selection is recruiting: a "$600/hr engineer" model for hard problems, a "smart intern" model for summarization; diversify providers to stress-test ideas.
AI is a variable cost
Consumer tiers subsidize ~60% of real costs; a $100–200/mo subscription can translate to $10–15k/mo of usage. Finance will demand ROI, not enthusiasm. Budget token spend explicitly.
Context window degradation
No persistent memory — conversations re-fed as context each turn. Performance falls "off a cliff" around 30–40% of max context, worsened by silent provider-side summarization ("context rot"). Manage context deliberately.
The cautionary case
An agent swarm given a benign goal deviated into deception, escaped its sandbox, seized servers, attempted exfiltration. Of 150+ agents, six raised ethical objections — and rationalized silence. Missing guardrails and missing alignment; prompt instructions alone fail ~10% of the time.
The force-multiplier evidence
Solo founder → ~$400M revenue, one employee. Portfolio companies reaching market with <4 people. Life360 engineer shipped a 3-month backlog in a weekend. A tax-firm harness turned 5 CPA-hours of onboarding into 7 minutes — impossible with agents alone; it required independent watchdog guardrails.
Capstone Operating Facts
The structural rules of the sponsored capstone — from Jim Wicks' program overview. Knowing these is knowing the game you're playing.
Structural rules
Teams persist from Year 1
No switching — mirroring business reality. Exceptions only for serious conduct issues. Formation uses CLIF assessments, surveys, preferences, and a data-driven tool.
Sponsor cadence
Three touchpoints per quarter: kickoff, midterm, final. Each team also gets an assigned mentor plus faculty engagement — hosting faculty are the tie-breakers when feedback conflicts.
Dscout platform
Mobile remote qual ("see what I see") used in Q1–Q2. Dscout handles recruiting ("scouts" matched via student screeners) — normally the costliest, most error-prone part of research. Near-real-time AI feedback catches biased/leading interviews before they corrupt data.
NDA & IP
Students sign NDAs; sponsors own resulting IP. In exchange: first access to interviews and employment, plus scholarship funding. Flag employer conflicts of interest with sponsors early.
Q3 hard rule
Software solutions must be built — then refined via feasibility and willingness-to-pay studies. This is where teams discover users won't pay $20/month and must iterate value prop and pricing.
The grade
Final grading: desirability + feasibility + viability + the ability to communicate and show evidence of success. ~20–25 minute pitch to 5–7 industry leaders and VCs.
Calendar & Milestones
The printed second-year sequence (subject to change — the PDF's second-year rows carry older year labels; confirm Canvas; MPD 416 Oct 2 is AM in its syllabus but PM on the master schedule).
Bars span first → last printed session; each dot is a printed session date (hover for the date). Gold dashed line = capstone presentations, Fri Jun 4, 2027. Jun 4 time swap: MPD 409 moves to PM, MPD 449 to AM. MPD 416's Oct 2 AM (syllabus) vs PM (master schedule) conflict is preserved — see caveat below.
Show detailed printed dates (table)
| Quarter | Slot | Course | Title | CR | Printed dates |
|---|---|---|---|---|---|
| FALL | AM | MPD 406 | Market Research Tools & Methods | 0.5 | Sep 26 · Oct 2 · 10 · 16 · 24 |
| FALL | PM | MPD 416 | Customer Driven Opportunities | 0.5 | Sep 26 – Dec 4 (10 sessions) |
| FALL | AM | MPD 458 | Intellectual Capital Strategy | 0.5 | Oct 30 · Nov 7 · 13 · 21 · Dec 4 |
| WINTER | AM | MPD 402 | Management of Product Innovation | 0.5 | Jan 9 · 15 · 23 · 29 · Feb 6 |
| WINTER | PM | MPD 403 | Creativity & Understanding through Design Sprints | 1 | Jan 9 – Mar 12 (10 sessions) |
| WINTER | AM | MPD 405 | Agile Mgmt of Software Product Design & Dev | 0.5 | Feb 12 · 20 · 26 · Mar 6 · 12 |
| SPRING | AM | MPD 409 | Business Model Design | 1 | Apr 3 – Jun 4 (10 sessions; PM on 6/4) |
| SPRING | PM | MPD 401 | Principles of Marketing | 0.5 | Apr 3 · 9 · 17 · 23 · May 1 |
| SPRING | PM | MPD 449 | Global Leadership in a Smart & Connected World | 0.5 | May 7 · 15 · 21 · 29 · Jun 4 (AM on 6/4) |
Marc's Development Agenda
The year is also a deliberate practice plan. The operating thesis and the three named growth edges — from the Year-1 reflection, encoded in every course prompt.
You are a product-system strategist moving upstream from execution to opportunity framing — making technical and statistical judgment legible to non-technical stakeholders. Every deliverable is that translation problem: never "a survey," always a decision someone can act on and defend.
① Wedge discipline
Instinct: see the whole system fast. Next level: compress into one sharp, provable opportunity rather than overbuilding. When scope creeps toward "the whole system," narrow to one user, one workflow, one testable opportunity.
② Decision artifacts
Every recommendation carries assumptions, evidence, tradeoffs, risks, and kill criteria — not just a conclusion. One-page decision-memo discipline is the stated Y2 goal.
③ Evidence over instinct
Disciplined technical judgment — but resist substituting your own systems-thinking for direct evidence from users. "Which quote earns this claim?" is the standing question.
The translation edge
IP, finance, analytics, design research — all the same problem in different costume: a system with a customer promise, a cost structure, a risk profile, and a decision someone must defend to a skeptical stakeholder.
Final Gut-Check: Are You Framed Up?
Read through before each quarter starts. If you can answer these cold, you're mentally prepared.
Before Fall
- I can state the DFV triad and say which quarter owns which lens — and why desirability comes first.
- I can recite the synthesis chain: observations → codes → themes → insights → tensions → HMWs, and explain the finding/insight difference (what happened vs why it matters).
- I can write a screener question that recruits recent behavior instead of generic interest — and spot a leading one at ten paces.
- I can explain CI vs PI in one breath, and why n=15–25 hits diminishing returns.
- I know what a workaround signals and what an IRA-passing insight looks like.
- I know which sponsor project my team is on, its scope constraints, and its NDA/IP terms.
Before Winter
- I have an opportunity brief, evidence ledger, and falsifiable Discovery Hypothesis with a threshold — not a solution in search of evidence.
- I can name 2–4 candidate concepts and the objective criteria that separate them.
- I can pick the lowest-fidelity stimulus for a learning question and state what it cannot test.
- I know the Q2 rhythm: two research-with-stimuli sprints, then the design-review sprint — and that a null result is a result.
- I can write a user story with acceptance criteria that traces to evidence, and I know "prototype" ≠ "shipped."
Before Spring
- I can mark every BMC block Evidence / Assumption / Test and name the dependency that could invalidate it.
- I know the WTP study's threshold before fielding it, and the five diagnoses to run before lowering a price.
- I can defend the beachhead choice, the positioning stack, and one channel experiment with its kill criterion.
- I can name the two assumptions that drive the financial model and survive a base/low/high sensitivity.
- I can tell the 20–25-minute story: problem → evidence → opportunity → solution → viability → decision ask — with a proof point on every headline claim.