GMI walkthrough notes

Status

Authored: 2026-05-12
Stage: Discovery / problem validation
Source: informal chats with Greg (GMI co-founder, also a co-lead on this product). Not yet a structured workflow walkthrough call.
Provisional. Items below are tagged F (fact: Greg's direct account or independently verifiable), A (assumption: derived but not yet confirmed), H (hypothesis: extrapolation about the wider market).

Use this as the raw log; structured outputs (workflow & pain map, supplier participation model, taxonomy notes) should cite back here.

Conversations captured here

2026-05-12 — informal chat covering supplier data shape, GMI's enrichment workflow, supplier-side asymmetry observation, MPN/SKU practice, APIC and Australian Music Association leads.

GMI's source-of-truth and workflow

(F) GMI uses Shopify as their central source of truth for product data. Justified primarily by data quality reasons (it's where they end up after enrichment), and incidentally because their retail D2C channel runs on the same Shopify store.
(F) Supplier feeds to GMI are typically minimal: CSVs containing MPN, MAP, RRP, and a few other commercial fields. No enrichment content (descriptions, images, specs).
(F) For enrichment content, the international brands tell GMI to retrieve it from the brand's own website.
(F) Greg manually builds additional CSVs by extracting images, descriptions, and other enrichment data from each brand's website.
(F) Greg combines the fragmented CSVs (commercial data from supplier + scraped enrichment) by importing them in sequence into Shopify, which becomes the merged record.
(A) This is a per-product, semi-manual process whose cost scales roughly linearly with catalog size. Not yet confirmed in operational terms (hours/SKU, error rate, frequency).
(A) The Shopify export shared on 2026-05-12 (research/inputs/gmi-shopify-export-2026-05-12.csv) is the output of this process. It is not the raw supplier feeds, nor a record of the work between raw and clean.

Supplier asymmetry observation

(F) From Greg: international boutique brands (GMI's direct suppliers) do not provide enriched data — only pricelists. Enrichment burden falls on the importer.
(F) At least some larger international suppliers operate their own dealer portals exposing rich product data. Confirmed examples: Fender (https://dealer.fender.com), Roland AU (https://dealer.rolandcorp.com.au). The APIC database assessment further supports this by showing those portals expose far richer fields than the same suppliers submit into central systems (Fender dealer portal: 166 spec columns; Roland: 26 columns; APIC: 1 free-text field). See ../landscape/apic.md.
(H) GMI's specific case highlights a potential use case for small suppliers without internal master-data infrastructure, where a central catalogue would be their source-of-truth (and the integration problem is data-creation tooling + enrichment workflow, not data transfer + normalization).
(H) Whether the broader AU music supplier landscape divides cleanly into "large with own infra" vs "small without" — or whether it's a spectrum, or has further distinct patterns we haven't observed yet — is unknown. The two observations above show at least two visible patterns; a clean two-bucket model is one interpretation among several and should not be treated as established.
(H) Regardless of how many supplier patterns exist, the supplier participation model probably needs more than one onboarding path. Post-MVP scoping territory; do not design now.

Identifier practice

(F) MPNs are common from manufacturers and are present in supplier feeds.
(F) GMI creates its own internal SKUs and uses them in Shopify (i.e. retailer-local identifiers distinct from manufacturer identifiers).
(H) This pattern is industry-typical: manufacturers issue MPNs (often GTINs too), retailers issue their own SKUs. For a cross-retailer product catalog this implies:
Identity resolution probably anchors on MPN (or GTIN) where available.
Products without an MPN (e.g. some boutique items) need a different resolution strategy.
Retailer-local SKUs are mappings, not the canonical identifier.
(A) This is a well-known pattern in catalog/PIM systems and is not novel — but Greg's account is a useful confirmation for the AU music case specifically.

APIC — covered in its own doc: ../landscape/apic.md. Greg has direct knowledge and access; GMI is in active conversation with APIC as a supplier.
Australian Music Association (https://australianmusic.asn.au). Greg has a relationship with the owners. Useful as a future channel for supplier- and retailer-side intel across the AU music space. Not yet engaged.

Open questions for the walkthrough

These need answers (from Greg directly, or via a structured workflow call) before downstream artifacts like the workflow & pain map can be authored confidently:

Approximate hours/week GMI spends on enrichment work, and how that scales with new-product cadence.
How often supplier price/stock updates arrive, and what triggers an enrichment+import cycle into Shopify.
Whether the manual enrichment work has ever caused customer-facing errors, and how those are caught.
How GMI categorizes products internally (current taxonomy, even if informal).
Whether any suppliers send images/descriptions in any form (PDFs, marketing emails, asset packs) — useful as raw input samples.
Is the two-bucket supplier asymmetry framing complete, or are there sub-patterns we're missing (e.g. distributor-mediated suppliers, multi-brand importers)?