Status note (2026-05-08): This plan was authored before the project's legal posture was established. The recruitment + retailer/supplier interview sprint described below is deferred until post-employment per ../constraints-and-separation.md. The risk-ranked assumptions, falsification thresholds, and commit/pivot/kill decision rule remain valid framing. Customer zero is now named — see customer-zero-profile.md. The during-employment validation path runs solely through the GMI design-partner relationship.

Customer Discovery Plan

Status

Stage: Problem validation
Authored: 2026-05-06
Owner: TBD (must be named before kickoff)
Provisional. Revise as evidence comes in.

Purpose

Convert the assumptions and hypotheses log into a validation program with falsification criteria, recruitment targets, and a decision rule. Produce evidence sufficient to commit to, pivot, or kill the proposed direction.

What this plan does

Ranks assumptions by load-bearing risk.
Specifies the cheapest test for each.
Defines what would falsify each one.
Sets recruitment targets, sample sizes, and a timeline.
Defines a commit / pivot / kill decision rule at the end of the sprint.

What this plan does not do

It does not validate solution shape, taxonomy depth, ingestion model, API surface, or pricing tiers. Those are post-validation artifacts.
It does not produce a roadmap.
It does not commit to MVP scope.

Risk-ranked assumptions

Ranked by load-bearing risk: if this is false, how much of the product thesis collapses?

A1. Retailers experience catalog pain that is severe, frequent, and currently funded

Severity: Critical. If false, no business.
Source: assumption log A1.
What "true" looks like: 8 of 12 interviewed retailers can name a person, hours/week, or dollar cost they spend on supplier-data wrangling, and at least 4 have actively tried to fix it (built spreadsheets, hired a junior staffer, bought a tool, paid an agency).
What would falsify it: Most retailers describe it as annoying but not costly enough to fund. No prior workarounds. No willingness to dedicate even a discovery hour.
Falsification threshold: Fewer than 4 of 12 show active funded behavior (time, money, or workaround already in place).

A2. The pain is consistent enough across retailers that one normalized layer is genuinely shared infrastructure

Severity: Critical. If false, this becomes a bespoke services business, not a platform.
Source: derived. Not in current assumptions log — should be added.
What "true" looks like: Across 12 retailers, the top 5–8 supplier sources overlap heavily (>60% supplier overlap across the segment), and the fields they care most about overlap heavily.
What would falsify it: Each retailer has a long-tail supplier mix with little overlap, or each has idiosyncratic field needs that defeat normalization.
Falsification threshold: Less than 50% supplier overlap across the target segment, or no consistent top 5 fields.

A3. There is a willing customer zero who will pay (or pre-commit) before the product exists

Severity: Critical. If false, MVP scoping is theoretical.
Source: derived from open-question #1.
What "true" looks like: At least 1 named retailer signs a design-partner agreement (paid pilot, LOI with target spend, or prepayment) within the discovery sprint.
What would falsify it: Strong verbal interest from many, written commitment from none.
Falsification threshold: Zero design-partner agreements after 12 interviews and explicit asks.

A4. Suppliers will participate when retailer demand exists

Severity: High. If false, the data side breaks. But mitigable — retailer scraping/manual ingestion can bridge early.
Source: assumption log + legacy precedent (MusiPOS/APIC).
What "true" looks like: 4 of 6 interviewed suppliers say "yes, if a retailer asks us to send to your platform, we will," and at least 2 will commit in writing.
What would falsify it: Suppliers actively block, citing channel control, exclusivity, or lack of resources.
Falsification threshold: Fewer than 2 of 6 willing to commit even informally.

A5. The first integration use case is concrete enough to define a minimal API

Severity: High. If false, "minimal API" stays vague forever.
Source: derived. Open-question #8.
What "true" looks like: At least 2 retailers can name the exact downstream system (POS X, ecom platform Y) and the exact catalog operation (initial load, daily sync, new product onboarding) they want first.
What would falsify it: Retailers can describe pain but cannot name the integration shape they would actually use first.
Falsification threshold: No retailer can articulate a concrete first integration without prompting from us.

Deferred (not validated in this sprint)

API-as-premium pricing assumption: too early. Validate after a paid pilot exists.
"MVP can deliver value before deep attribute normalization": this is a scoping decision, not a discovery one. Defer to MVP scope memo.
Mid/small retailers adopt faster than large: useful segmentation hint, but not load-bearing. Treat as recruitment heuristic, not a tested hypothesis.

Research methods

Primary

Retailer interviews (semi-structured, 45–60 min, video). Mom Test discipline: ask about past behavior, not future intent.
Artifact review during interview: ask retailer to walk through their last supplier-data update in real time (screen-share if possible). This is the single highest-signal activity.

Secondary

Supplier interviews (30–45 min, video).
Existing-tool teardown: 1–2 hours per relevant adjacent tool (legacy MusiPOS/APIC behavior, generic PIM tools, retailer ERPs). Used to sharpen "why existing solutions are insufficient."

Explicitly out of scope

Surveys. Low signal for this question.
Landing-page smoke tests. Premature; we don't have positioning yet.
Building a prototype. Premature.

Target retailer profile (recruitment heuristic)

Hypothesised, not validated. Used to seed the recruitment list, not to constrain learning.

Australian music retail (instruments, audio, related accessories).
1–10 stores; some ecommerce presence.
20+ active suppliers in their catalog.
Owner/operator or small leadership team (decision speed).
Has felt pain at least once: a supplier price update, a new supplier onboarding, a POS-to-ecom drift incident.

Disqualifiers (do not put at top of list): - Largest retailers with mature in-house data ops. - Single-supplier or single-brand retailers (pain is too narrow). - Pure dropshippers (different problem shape).

Recruitment plan

Target list size: 25–30 retailers identified, 12 interviewed.
Channels: industry associations, supplier referrals, owner-operator networks, LinkedIn outreach, warm intros from anyone in your network.
Incentive: founder access, influence over taxonomy and roadmap, optional gift card. Avoid cash for interviews — biases responses.
Recruitment is usually the bottleneck. Start day one. If after week 2 you have fewer than 6 booked, the plan is failing and recruitment becomes the priority, not the interviews.

Supplier targets

6 suppliers, mixed by size and current data sophistication.
At least 2 with prior MusiPOS/APIC participation.
At least 2 with no prior central-data participation.

Sample sizes and timeline

Stream	Target	Notes
Retailer interviews	12	Saturation usually around 8–12 in a single segment
Supplier interviews	6	Secondary stream, lighter cadence
Design-partner conversations	3–5	Subset of retailers showing strongest signal

Sprint length: 6 weeks.

Week	Activity
1	Recruitment, finalise discussion guides, first 2 retailer interviews
2	4 retailer interviews, 1 supplier interview, weekly synthesis
3	4 retailer interviews, 2 supplier interviews, midpoint review
4	2 retailer interviews, 3 supplier interviews, design-partner conversations begin
5	Follow-ups, design-partner offers, gap-fill interviews
6	Synthesis, decision memo, go/pivot/kill

Cap: do not extend past 8 weeks. If the answer isn't visible by then, the answer is "not yet."

Retailer discussion guide (semi-structured)

Open with rapport, not the product. Never describe the product unless explicitly asked at the end.

1. Their world

Walk me through your business: store count, supplier count, channels.
Who handles your product data day to day?

2. Concrete recent event

The last time you onboarded a new supplier or did a major catalog update — walk me through what actually happened.
(Screen-share request) Can you show me the last spreadsheet, email, or file you got from a supplier?
How long did it take? Who did it? What broke?

3. Pain frequency and cost

How often does this happen?
What does it cost you in time or money in a typical month?
What's the worst version of this you've had in the last year?

4. Current alternatives

What have you tried to fix this? Spreadsheets, staff, agencies, tools?
What did you pay for?
Why did it stop working, or why is it still in place?

5. Workarounds and adjacent spend

Are there things you'd love to do that you currently can't, because of data quality?
Have you ever paid for a tool that touched this problem?

6. Integrations

What systems would the data need to land in for it to be useful — POS, ecom, ERP, accounting?
Which one matters first?

7. Closing

If a service existed that gave you clean, normalized supplier data already mapped to your categories, who at your company would care most, and how would you decide whether to use it?
Can we come back to you in a few weeks with a concrete proposal?

Do-not-ask

"Would you pay for X?" — meaningless.
"Would this be valuable?" — meaningless.
Leading questions about features.

Supplier discussion guide (semi-structured)

1. Their world

How do you currently distribute product data to retailers?
How many retailers do you serve in AU? Which channels?

2. Concrete recent event

Last time a retailer asked for a catalog update — walk me through what happened.
What format did you send? Why that format?

3. Pain and cost

How much time/effort goes into retailer-facing data?
Where does it break?

4. Centralization willingness

If a retailer asked you to send data once to a shared platform that retailers pulled from, would you do it? Why or why not?
(For those with MusiPOS/APIC history) What worked, what didn't, what would you not repeat?

5. Constraints

Channel control concerns?
Pricing visibility concerns?
Resourcing constraints on your end?

6. Closing

If a retailer you already supply asked you to participate, would you?
Could we come back with a specific ask in a few weeks?

Synthesis cadence

Weekly 30-min synthesis: review tagged interview notes, update assumption status (supported / weakened / falsified / undetermined), update open-questions log.
Midpoint (end of week 3) review: are we on track to hit thresholds? If not, recruit harder or narrow segment.
Final synthesis: decision memo using templates/decision-memo-template.md.

Note-taking standard

Verbatim quotes preserved.
Behavior > opinion. Tag each insight as behavior (something they did) or opinion (something they said they would do). Behavior outweighs opinion in synthesis.

Decision rule (commit / pivot / kill)

Evaluate at end of week 6.

Commit (proceed to MVP scope memo)

All of: - A1, A2, A3 meet their thresholds. - At least one named design partner with written commitment. - At least one concrete first-integration use case named by 2+ retailers.

Pivot (re-scope, not re-validate from scratch)

Any of: - A1 holds but A2 fails → reconsider as bespoke services or per-retailer onboarding, not platform. - A1 and A2 hold but A3 fails → segment was wrong; redo recruitment for a different retailer profile, max 4 weeks. - A4 fails → consider retailer-side ingestion path; suppliers become a later concern.

Kill

A1 fails. The pain is not funded. Without funded pain, none of this matters.

Outputs of this sprint

Updated assumptions-and-hypotheses.md with status per assumption.
New retailer-workflow-and-pain-map.md (current-state document).
New customer-zero-target-profile.md (refined ICP based on evidence).
Decision memo: commit / pivot / kill.
Optional: design-partner agreement(s).

Open questions before kickoff

These must be resolved before week 1 begins: - Who owns this sprint end-to-end? - Who conducts the interviews? (Founder presence in B2B discovery is high-leverage; do not delegate to a researcher who can't make commitments.) - What is the recruitment-channel plan, concretely? Names, intros, lists. - Is there budget for a part-time recruiter or coordinator if recruitment lags? - What is the minimum commitment we are willing to offer a design partner, and what do we expect in return?

Suggested next step

Resolve the five open questions above this week. Begin recruitment in parallel — recruitment lead time is typically 1–2 weeks and is the rate-limiting step.