For a marketing operations leader preparing a desk-organizer campaign, this note proposes one internal task: put the exact creative beside the records offered to support it. Give the product owner and internal wording reviewer a comparison they can inspect, including the parts that do not match.
The marketing-claims coordinator described here is a MAJLS proposed operating design. It is not a demonstrated MAJLS product capability or a report of client results. In this design, a Virtual Employee prepares the comparison within authorized internal records; people assess the records, decide wording and authorize release. The model never certifies truth.
Borrow a provenance vocabulary, not a ready-made workflow
W3C's PROV Model Primer describes the provenance of digital objects as their origins and provenance records as descriptions of entities and activities involved in producing or influencing an object.[1] PROV-DM defines an entity as a thing with some fixed aspects, an activity as something that acts upon or with entities over time, and an agent as something bearing responsibility for an activity or entity.[2]
For this proposed register, treat the creative version and source record as the things to identify, comparison preparation as the activity, and each person's contribution as a named responsibility. Record who supplied the evidence separately from who reviewed the wording. That mapping is MAJLS's adaptation.
PROV-DM is domain-agnostic and allows domain-specific information to be included.[2] The Primer also says roles are application specific; PROV does not define particular roles.[1] These sources supply a vocabulary for recording origins and responsibility, not a marketing approval procedure. Our fields, stop conditions and human decisions below are design choices. A documented origin does not, in this workflow, clear a claim for use.
MAJLS process map: creative to internal review
- Freeze the context. Capture creative version, channel, audience, product variant, exact wording, image brief and evidence cutoff. Retain nearby copy and the complete creative snapshot. Keep proposed edits separate from the input.
- Extract candidate claims. Preserve literal wording. Record a possible reading of an image as a question for the reviewer, not as an established audience response.
- Compare scope. Link each row to a retrievable record version and exact passage or observation. Include the method, configuration, limitations, record supplier and technical owner. A similar product name is not a match under this proposed rule.
- Prepare the handoff. Ask the product owner to confirm the record and its relevance. Ask the internal wording reviewer to assess the comparison and choose wording for the specified creative. Leave either unanswered request pending.
Use states that describe preparation: evidence collected, reviewer needed, held for gap, and human wording decision recorded. Require a person's attributable decision for the last state. Keep release authorization in a separate field.
Filled example: a notebook-capacity campaign
Everything below is illustrative: the product, records, dimensions, observations, names and times are synthetic. No real product was tested and no campaign was approved. The packet's input snapshots are filled fictional records, not links to real test files.
The fictional Alder Desk Tray campaign uses creative ADT-v03. Its draft product page says, "Holds all your work notebooks," beside image IMG-v03. The image brief specifies a larger notebook. The fit record describes only two smaller notebooks together.
| Field | Filled value |
|---|---|
| Product and variant | Alder Desk Tray, narrow slot, N1 |
| Creative and audience | ADT-v03; IMG-v03; draft product page for office-accessory shoppers |
| Evidence cutoff | 1 October 2026, 09:00 UTC; synthetic review clock |
| Record and locator | FIT-N1-07 v1; observation O1 and limitations L1 in the packet's synthetic input snapshots |
| Method | Manual insertion and removal of two 148 × 210 × 12 mm notebooks upright together in each of two N1 specimens |
| O1 | "The two specified notebooks fitted together without forced insertion in each specimen." |
| L1 | "Only the specified configuration was observed; larger formats, mixed thicknesses, load, durability and other variants were not tested." |
| Record responsibility | Lena, fictional product owner and record supplier; technical confirmation pending |
Claim-to-evidence register
| Claim and creative locator | Comparison and unresolved gap | Requested human action | State |
|---|---|---|---|
| C01: ADT-v03 headline, "Holds all your work notebooks." | O1 records two specified notebooks in N1. L1 leaves other sizes and quantities outside the observation. | Lena confirms the record and configuration; Priya, internal wording reviewer, assesses the broad wording. | Held for gap |
| C02: IMG-v03 brief, notebook 210 × 297 × 18 mm | No observation for that configuration appears in the synthetic inputs. The brief describes the image; it is not a fit record. | Lena checks configuration; Priya assesses whether the image suggests a fit the record does not address. | Held for gap |
Both rows use the same variant, creative version and cutoff. C02 is a review question. The coordinator must not promote it to a finding about what shoppers believe. A narrower headline alone would leave that row unanswered.
The proposed wording and handoff
Prepare this option without applying it: "Fits two notebooks measuring 148 × 210 × 12 mm each, upright together in the N1 slot." Lena still needs to confirm the technical record. Priya needs to decide whether that candidate accurately communicates its limits alongside the proposed image. Neither decision is supplied by a model's fluent summary.
Packet RR-ADT-01 contains:
- ADT-v03, IMG-v03 and FIT-N1-07 v1 as labeled synthetic input snapshots, plus both claim rows.
- O1, L1, the method and Lena's pending technical-confirmation request.
- The narrower wording and an image option showing only the recorded configuration; neither is applied or approved.
- Priya's decision block: pending; creative covered none; conditions unresolved; decision time absent.
- Omar's separate release block: authorization absent; publication blocked.
At the synthetic 09:20 UTC handoff, place the packet in the internal review queue with Lena's record question and Priya's wording question intact. For consequential publication, named human release owner Omar retains approval after the internal review decisions. The Virtual Employee may not waive gaps, change the campaign, publish it or contact an external audience.
Keep the decision attached to the version
The PROV Primer treats the result of each revision as a new entity.[1] Our operating rule borrows that distinction: if the wording, image, evidence or product variant changes, prepare a new comparison and identify its relationship to the old one. Do not carry a v03 decision into v04 silently.
Record the human decision's author, role, exact creative and evidence versions, scope, conditions and time. Pause the handoff when evidence is inaccessible, variants conflict or a required answer is missing. Use those observable gaps to route the question to its owner; model confidence cannot fill an absent record or decision.
Start with one creative snapshot. Fill one row per claim, attach the relevant record passages and leave every unresolved difference visible for the next person.