THE BRIEF
Build, test, improve, explain.
Build a no-code prototype for an entrepreneurial opportunity. Explain why you chose it, show what you tested, and use the findings to improve it. Work on your team project; each student submits their own evidence and reflection.
20% participation + 80% portfolio. At least 5.5 in each. Prompt points do not change the grade. The advertisement is required, without additional weighting. Deadlines and exam arrangements are announced in Google Classroom.
EIGHT REQUIRED ITEMS
What should the reviewer be able to verify?
| Portfolio item | Evidence to prepare | Question to check |
|---|---|---|
| 1. Opportunity | Specific audience, problem, and supporting observations. | What is known, and what remains an assumption? |
| 2. Concept | Why this solution and prototype scope make sense; connect to entrepreneurial theory. | Which important assumption does the prototype test? |
| 3. Working prototype | Live website and matching GitHub repository links. | Can someone complete one useful task? |
| 4. Testing | Participant context, task, method, actual observations, and interpretation. | Could someone distinguish evidence from your opinion? |
| 5. Iterations | Before, change, reason, and check or clearly labelled planned retest. | What evidence caused this decision? |
| 6. Tools | Choices, alternatives, tradeoffs, and why the tools fit the task. | Why were these suitable for this prototype? |
| 7. Individual reflection | Your contribution, learning, changed assumptions, and next step. | What did you personally do and learn? |
| 8. Advertisement | Creative link and copy; audience, value promise, and call to action. | Does the landing page deliver the promise? |
WORKED EXAMPLE / FICTIONAL
From a vague claim to useful evidence
Weak: “People liked our website, so we improved it.”
More useful: “I asked three volunteer shoppers to submit a stock enquiry. Two could not find the action. I moved it next to the product details and added a confirmation message. I will retest whether new users can complete the task and explain what happens next.”
The participant count and observations above are invented for teaching. Notice that the retest is planned, not presented as completed. Replace every example with your actual work.
Your turn: choose one claim in your draft. Write the observation, interpretation, decision, and next check separately.
EXERCISE 1 / CUSTOMER + OFFER
Be specific enough to find someone.
- Our intended user is… The person who pays is…
- The problem occurs when… Our evidence is…
- The riskiest assumption is…
- For [audience], we offer [outcome] through [solution].
- The first step we invite them to take is…
Choose a research invitation, test, enquiry, or pilot that honestly matches your current stage.
EXERCISE 2 / FIND + QUALIFY
Five useful records beat an unexplained list.
Find five relevant candidates, or document an opt-in recruitment route for a consumer project. Record a source, observable evidence of fit, and what remains unknown. Five is a classroom practice target, not an official submission quota.
Qualification questions: does this match our segment? Is there evidence of the problem or only a guess? Is there an appropriate way to invite a conversation?
Copyable AI research prompt
Our offer is [offer] for [segment]. Based only on [sources], return the source URL, observed fact supporting fit, uncertainty, and next question. Do not invent contacts, email addresses, budgets, needs, or buying intent. Mark missing facts unknown. Separate your interpretation from the source evidence.
Check each AI-produced claim against its source. Keep real contact data out of public repositories, course notes, and shared slides.
EXERCISE 3 / CRM
Make the next action obvious.
Use the Attio lab for a CRM walkthrough, or import the blank CSV into a spreadsheet you already use. The sample row is fictional and must not be treated as a researched lead.
Pipeline: identified → reviewed for fit → contacted → replied → discovery/test booked → offer/pilot → won/lost. Allow “not a fit” and “do not contact” from any stage. For validation, use “test completed” and “learning recorded” instead of closing a sale.
- Record identity, source, date checked, evidence, and unknowns.
- Choose the stage from an event that actually happened.
- Assign one owner and a dated next action.
- Record outcomes and stop follow-ups when someone declines.
Peer check: give the table to a teammate. Can they tell what happened and what should happen next?
EXERCISE 4 / CONVERSATION
Offer a useful next step.
First-message template: “Hi [name], I noticed [verified observation]. We are testing [specific solution] for [audience]. Is [problem] something you encounter? If so, would a short conversation be useful?”
Review three drafts. Remove false familiarity, unsupported claims, and pressure. Sending during class is optional. Public contact information alone does not establish permission for marketing; use an appropriate channel and check applicable requirements before live outreach.
Discovery: when did this last happen? What did you do? What was difficult? What would make a different approach worth trying?
Follow-up: log the reply and agreed next action. For this exercise, use at most one unanswered follow-up, then close the task. Stop on refusal.
Copyable AI message-review prompt
Review this draft against these verified facts: [facts]. Identify unsupported claims, irrelevant personalisation, pressure, and an unclear next action. Suggest a shorter draft for human review. Do not send it.
Copyable role-play prompt
Play a potential customer from [segment]. Let me ask about your current workaround for [problem]. Do not immediately agree with my idea. Afterwards, point out my leading questions and unsupported conclusions. Treat this as simulated practice, not validation evidence.
EXERCISE 5 / CAPTURE TEST
Follow the enquiry all the way through.
Ad or website → clear action → form → saved record → owner → next action.
A manual handoff is acceptable if you document and demonstrate it. An integration is working only when you can show the saved result.
| Case | What to observe | Record |
|---|---|---|
| Valid dummy enquiry | Visitor confirmation, saved record, owner | Actual result + evidence |
| Required information missing | Clear error and recovery | Actual result + evidence |
| Duplicate | Can the team recognise and handle it? | Actual result + evidence |
| Failed handoff | Visible failure and a recovery path | Actual result + evidence |
SEVEN-DAY EXPERIMENT
Write the decision rule first.
“We believe [segment] experiences [problem]. We will test [offer] through [channel]. If we observe [signal], we will [decision]. Otherwise, we will investigate [alternative].”
Suggested practice: ten reviewed candidates or an opt-in route; five appropriate invitations where permitted. Adapt to the project. These targets do not add grading requirements or guarantee clients.
Record researched, contacted, replied, booked, tested, offers made, and won separately. Show denominators and dates. Five invitations and two replies is 2/5 replies, not proof that 40% of the whole market wants your product.
Bring back: an anonymised extract, actual results, one customer insight, one justified change, and your individual reflection.
PORTFOLIO CLINIC
Connect the exercise to the existing fields.
- Opportunity/concept: customer research and the assumption you tested.
- Testing: actual conversations, prototype tasks, capture checks, and their results.
- Iterations: what changed because of that evidence.
- Tools: why the table, form, AI tool, or integration suited the task.
- Reflection: your personal work, judgement, and next question.
A CRM screenshot by itself is not proof of demand. A role-play is practice. An unanswered message is a recorded outcome, but does not establish why someone did not respond.
QUESTIONS STUDENTS OFTEN HAVE
Before you submit
Do I need to make a sale to pass?
The current brief assesses your prototype, entrepreneurial reasoning, testing, iteration, and reflection. It does not require a sale. Show honest evidence and learning.
Can the team use the same prototype?
Yes. Work on the team project, while each student submits their own evidence and reflection and identifies their contribution.
Can AI-generated feedback count as user testing?
Label it as simulated practice or a prompt experiment. It cannot replace observations from real users or peers when claiming actual validation.
Is the CRM a new required portfolio item?
No. Use it to support the existing testing and iteration evidence. The eight required items remain the same.
How do I know if my evidence is strong enough?
Use the official rubric in the hub. Make the reasoning traceable: assumption, test, observation, decision, change, and check. A reviewer should be able to open the relevant links and understand what you personally contributed.
What if nobody responds?
Record the actual outcome and examine possible explanations. Try a justified change to your segment, invitation, or channel. Do not invent replies or treat silence as proof of one explanation.
EXIT CHECK
Leave with a named next action.
- Which of the eight items is weakest?
- What concrete evidence would improve it?
- Who owns the next action, and when will it happen?
- Have you saved your actual work in the portfolio?