Learn and decide: what the evidence says and what you’ll do next
Early evidence is small and messy. Tonight you read it honestly, make a decision you can defend, and plan your second loop.
By the end of tonight you can…
- Read early evidence with counts, thresholds, segments, quotes and a stated confidence
- Decide to persevere, iterate, pivot, pause or stop, and name the evidence behind it
- Place your product on the lifecycle and calculate your runway from your own numbers
- Plan how your team grows, and write a contractor brief whose result can be checked
- Present a five-minute demo and a one-page decision memo that plans loop 2
OUTPUT · Decision memo (with loop 2 plan) + five-minute demo
Reading early evidence: five moves, in order
- COUNTRaw numbers with their base: “5 of 10 tutors”, never “50%” on its own
- COMPARESet the result against the threshold you wrote before launch, as in week 2
- SEGMENTSplit by a meaningful group: who succeeded, who stalled, and what differs
- QUOTEAdd what people said and did, in their words, including the critics
- CONFIDENCESay how sure you are and why: sample size, time span, who took part
With ten users, one person moves the result by ten points. Lead with counts, then say how sure you are.
“Ten people liked it”
“Ten tutors tried TutorDesk and loved it. It’s working, so let’s add payments.”
Illustrative: 5 of 10 tutors confirmed a booking within 14 days; the threshold was 6. Tutors with 15+ learners: 4 of 5. Under 15: 1 of 5. Confidence low. Decision: iterate.
Ask
- Did TutorDesk meet its threshold, and by how much did it miss?
- Is 4 of 5 against 1 of 5 a finding, or a hunch to test?
- What would you need to see to be more confident?
- Why is “add payments” the wrong next step on this evidence?
Analyse your launch data, with honest uncertainty
I launched [MY PRODUCT] to [NUMBER] first users on [DATE]. I need to decide what to do next, and I’ll put your analysis in a decision memo, so it must not claim more than the data supports. My threshold, set before launch: [THRESHOLD]. <data>[PASTE EVENT COUNTS]</data> <notes>[PASTE NOTES AND QUOTES]</notes> For each funnel step, give the count with its base (e.g. “5 of 10”) and compare it with the threshold. Suggest one segment worth checking and show its counts. Rate your confidence (low, medium, high), stating the sample size and why. Where you’re unsure or the data can’t answer a question, say so and name what would settle it. Done means a short table, three plain sentences and a list of open questions.
Check the response
- Every number is shown with its base, e.g. 5 of 10
- Confidence is stated alongside the sample size
- Gaps are named rather than filled with guesses
Decide: persevere, iterate, pivot, pause or stop
- PERSEVEREThreshold met and the reasons are clear: keep going and test the next risk
- ITERATEClose, with a visible blocker: change one thing and measure again
- PIVOTThe problem is real but this solution or segment is not: change one big thing
- PAUSEToo little evidence, or something outside your control: set a date to look again
- STOPStrong evidence it isn’t worth solving for you: stop and record what you learned
Every decision names its evidence and the result that would change your mind. Stopping well is a success.
Return paths: where evidence sends you back
- MAP GATEPain not backed by real conversations: map again, or stop
- VALIDATE GATEGap or willingness to pay unproven: back to 02 Idea or 03 Ideation
- PLAN GATEObjectives without thresholds set before building: back to Validate
- ASSEMBLE GATEWeak grade: back to 07. Weak demo: back to 05, or 03 if the concept is wrong
- IGNITE GATEMust-pass failure: back to 07. Growth evidence starts the next loop at 01
Pivot sends you back to Map; iterate to Validate or Plan. Going back on evidence is progress.
The product lifecycle: where you are decides what you do
- INTRODUCTIONFew users, uncertain value. Learn fast: first value, early retention
- GROWTHDemand repeats. Strengthen onboarding, reliability and support as numbers rise
- MATURITYGrowth slows. Protect customers and margins; improve rather than reinvent
- DECLINENeeds or markets move on. Renew the offer, or retire it responsibly
A model, not a forecast: products can stall or skip stages. Every product in this room is in introduction.
Runway arithmetic (illustrative numbers)
“We have £18,000 saved, so we have plenty of time.”
£18,000 ÷ £1,500 a month = 12 months. Add a £3,000-a-month contractor: £18,000 ÷ £4,500 = 4 months. Twenty tutors paying £15 a month: £18,000 ÷ £1,200 = 15 months.
Ask
- Which single change moves TutorDesk’s runway the most?
- What must be true before a £3,000-a-month contractor is worth it?
- How many tutors paying £15 would cover the £1,500 monthly spend?
- What is your own runway, in months, from your own numbers?
How the team grows, one proven bottleneck at a time
- YOU AND CLAUDEYou own the product and decisions; Claude drafts, builds and reviews with you
- SPECIALISTSShort, bounded help where risk is high: security, accessibility, design, payments
- CORE TEAMLong-term people for product, engineering and design once demand repeats
- GROWTH ROLESSupport, marketing, sales and operations once growth needs dedicated owners
Hire to remove a bottleneck you can show, not to look bigger. Headcount is a cost, not a result.
Briefing a contractor: badly vs well
A weak brief
- “Make the booking page better, about £500”
- No link to users, evidence or the Product Book
- Done is whatever the contractor thinks is finished
- Shares your admin login and production secrets
A strong brief
- Outcome: tutors see new requests within a minute, on phone and laptop
- Context: the Product Book, CLAUDE.md and three user quotes
- Given/When/Then acceptance criteria, plus your Definition of Done
- Their own access, work via pull requests, no production secrets, IP agreed in writing
A contractor brief is a Claude briefing for a person: context, goal and why, examples, and what done looks like.
Loop 2: MVPAI again, at scale
- MAPWhy people stay or leave
- VALIDATEThe model and the AI bet
- PLANAI experience, platform
- ASSEMBLEAn AI slice, with evals
- IGNITESecure, sell and grow
Weeks 7–12 run MVPAI again with more users, more evidence and higher stakes.
MVPAI and the industry lifecycle: one journey, two vocabularies
MVPAI
- Map: the problem statement and the idea, backed by real conversations
- Validate: concepts scored, competitors, a business model and a willingness-to-pay test
- Plan, then Assemble: objectives and architecture, then build, demo and grade
- Ignite: productionize, launch and grow, then loop again from step 01
Words you will hear at work
- Discover: research the problem and the people who have it (Map)
- Shape: frame the opportunity and choose the bet (Validate)
- Design, Build: the experience, the system and the code (Plan, Assemble)
- Launch, Learn: release, measure and decide (Ignite)
Same journey, different labels. MVPAI adds a gate to every phase and must-pass checks that block.
Write your decision memo and prepare your demo
One page a sceptical reader would trust, and a five-minute demo that shows your first value moment live.
- Fill the funnel table: each event as a count with its base, against your launch threshold (10 min)
- Write the decision, the evidence for it, and the result that would change your mind (10 min)
- Add “What we don’t know yet”, your runway in months and loop 2’s first question (8 min)
- Script the demo: problem, live first value moment, evidence, decision, next step (10 min)
Decision memo in product/06-decision-memo.md + demo script
Capstone and the Ignite gate: demo and decision memo
- The demo runs the first value moment live on the deployed product, within five minutes
- Every number in the memo has its base and is compared with the threshold set before launch
- The decision is persevere, iterate, pivot, pause or stop, and is tied to named evidence
- The memo includes a segment, a quote and at least one piece of evidence against the decision
- A “What we don’t know yet” section states confidence and how loop 2 will find out
- The memo states the runway in months and the first question for loop 2
NEXT · Loop 2 starts in week 7 with Map: bring your memo’s first question and your first-users list.
© 2026 MVP Studios Ltd. MVPAI™ and the MVP Founders Programme materials are original works of MVP Studios Ltd. All rights reserved. No copying, redistribution or commercial use without written permission. Course terms