The structural fact of a co-manufacturer's traceability life, stated without decoration: you emit events for someone else's programme. The mandate arrives through the commercial relationship — a customer's quality addendum with a date on it — for a record whose value accrues to the customer, audited against you by the customer, on a clock the customer's regulators and retailers set. The Plant Quality Director who owns this is accountable for compliance without ownership, and the honest question is never whether to comply. It is in what architecture, because that choice decides whether the second mandate costs as much as the first.
What best-in-class looks like on the public record — and what it cost to be that anonymous
The best-documented supplier posture in the public record is GS1 US's Golden State Foods case study, published February 9, 2024: case-level RAIN RFID at the Opelika, Alabama protein plant, pre-tagged corrugate, GS1 Tag Data Standard 2 encoding carrying the production date, and a 100% encode success rate — GS1 US's figure, on GSF's line. Read it twice and notice the second fact: the study names the buyer only as "a leading QSR customer" and names no vendors at all. That is how this industry talks. The supplier does the work, hits the number, and stays anonymous about whose programme it serves — because the programme is not theirs. The case study is the shape of exactly the obligation on your desk: a customer specified the encoding, the grain, and the data element, and the plant's job was to execute at scale.
Reading a mandate: clause by clause, capability by capability
A mandate document decomposes into a small set of clause types, each mapping to a line or data capability. A model mapping — illustrative, but the clause types are the ones that recur:
| Mandate clause (customer's words) | What it actually requires of the plant |
|---|---|
| "Case-level identification per GS1 standards" | Serialized case marking — tag or 2D — with GTIN + serial in a standard encoding |
| "Production date / lot must be machine-readable" | The lot and date carried in the mark's encoding, not only on the ASN |
| "Shipment-level traceability data within 24 hours" | Shipping events emitted per case or pallet, linked to the ASN you already send |
| "Records available for audit / recall exercise" | A queryable event history, at lot grain, with retention |
| "Support our FSMA 204 KDE obligations" | Your shipping CTE data, sortable and electronic, mapped to their receiving CTE |
Nothing in that table is exotic. The trap is in the delivery clause — because there are two architectures for handing the customer their data, and they price wildly differently at mandate number two.
Join-their-network vs. grant-them-a-view
Architecture one: join each customer's system. Portal accounts, per-customer file formats, per-customer field mappings, per-customer audits of per-customer copies. The first integration is a project; every subsequent customer is another project; and your own record fragments into N partial copies, none of them yours. Switching cost compounds against you: five years in, you are operating five customers' compliance systems and own no asset at all.
Architecture two: emit once, grant views. The plant captures its events once — production, casing, shipping, at the grain the strictest mandate requires, each event carrying its who (the attested observer on the line or at the dock) distinct from capturedBy (the warrantor account) — and each customer receives a SharingGrant: a scoped view over exactly their SKUs, their POs, their shipments. On visibility.cloud the grant compiles down to spine-native scopes over one catalog; the customer sees their slice, the auditor re-derives it, and the record stays one record. The projection every customer reads validates as conformant EPCIS 2.0 against the pinned official schema — no conformance attestation has ever been issued, and none is claimed — which matters because conformant is what makes the same emission legible to every customer's system instead of one.
One record, N customers. Per-account portals multiply audits; per-account grants multiply nothing but read access. And the events you emit can carry the transaction context your customers already live in — the PO and the ASN are core CBV vocabulary, not custom fields — which is why the ASN you already send is the natural join point: ASN and PO context in traces.
Fund it to the account, not to an ROI fiction
The Director of Customer Programs will ask the only question that matters commercially: what is this worth? Answer it honestly — cost-to-serve, priced against the account. A co-manufacturer does not buy traceability for internal ROI; the internal-ROI spreadsheet is a fiction everyone in the room politely ignores. The real math is: this account carries this share of the book, its mandate has a date, and capability-once-grants-many means the next account's mandate costs a grant, not a project. That is also why capture itself is priced at zero on this platform — $0 per event, stated as intent, not terms, until terms are published — with revenue attached to answers: traces, custody evidence, grants, seats. A meter on capture would tax the completeness your customers' audits depend on.
Your customer's side of this decision — the brand owner choosing the mark and the event layer you will end up executing — is the pillar: Sunrise 2027 for brand owners. The bifurcation the same boundary creates inside the brand's four walls is the sibling read: running two traceability regimes in one plant network.
Where this goes next
visibility.cloud provisions capture workspaces from the seat list, in order. The way in is the interview: email first, under a one-message promise, then a short branching sequence about your plants, your customer mix, and the mandate on your desk — ending in a written read for your situation. The final step locks.
→ Start the interview — it is questions, not a demo.