The Northbridge Procurement Standard

Why verification before payment?

After publication, corrections to URL path, DOM disclosure or schema markup are no longer enforceable.

A publisher who has shipped an article under a /sponsored/ path will not change that path after invoice approval. The only mechanic that reliably bites is coupling the final invoice to evidence of fulfilled criteria. The verification workflow therefore runs before every final invoice, not after. Price coupling is the actual legal lever, not the advertorial contract.

The Procurement Standard is methodologically designed. It operationalises supplier due-diligence checking regardless of whether a regulatory obligation bites in the individual case. In a direct LLM contractual relationship, the check logic moves into the regulatory application scope. The precise classification of the relevant European and national interpretation lines sits in the DE annex of the Codex and in ongoing legal validation.

Two responsibility levels, two sanction mechanics

Three domain-policy criteria belong to the publisher pre-check. Fifteen criteria belong to briefing compliance.

The eighteen criteria distribute across two levels following different responsibilities. Quality publishers accept enforcement claims only in areas within the sphere of influence of the individual order. Domain-policy decisions such as robots.txt, paywall architecture or domain reputation are strategic publisher decisions outside the individual-order negotiation.

2.1Publisher pre-check before booking

Three criteria decide at the domain-policy level and apply independently of the individual article: A 03 domain reputation, A 05 paywall status, A 06 bot policy. They are go or no-go. If a publisher does not meet them, the article is not booked. Re-validation quarterly. No invoice consequence after booking, because these criteria are not part of the briefing frame.

2.2Briefing compliance before final invoice

Fifteen criteria sit within the publisher’s sphere of influence for the concrete article: A 01, A 02, A 04, A 07, A 08 plus B 01 to B 10. If the publisher honours the briefing, the final invoice is billed at the full classification per the price-factor model. Briefing FAILs are usually not repairable after publication and lead to invoice reduction.

The eight binary A-criteria

Eligibility filter. A single A-FAIL disqualifies the placement as a citation carrier.

The A-class checks whether a placement in the retrieval engines’ candidate set is at all indexable, crawlable and editorially classified. Every criterion is a binary filter: FAIL means disqualification, regardless of price, publisher name and content quality.

The methodologically central point of A 02 sits in separating textual inline disclosure from advertorial template. A retrieval-neutral disclosure marker below the headline satisfies both the legal disclosure duty and the eligibility requirement. A <div class="advertorial"> wrapper structure embedded in the article is visually identical in effect but engine-side disqualifying. Compliance and eligibility do not exclude each other; they are operationalised separately.

The ten gradual B-criteria

Lift levers. Every fulfilled B-criterion raises citation probability per engine, once the A-class is fully fulfilled.

The B-class determines citation lift once the A-class is fully fulfilled. All ten criteria belong to briefing compliance. They sit within the publisher’s sphere of influence for the individual article and are verified before the final invoice. The criteria act additively. Every additionally fulfilled criterion raises citation probability per engine, without any single criterion alone deciding eligibility.

At B 01, editorial final responsibility is to be separated from the authorship question. A narrow reading ("actual editor, no guest article") would conflict with Press Code Item 7. The standard’s reading mirrors actual publisher practice: principal or agency preparatory work passes through editorial review, revision and substance validation by the bylined author, who carries the final responsibility.

The two-stage verification workflow

Stage 01 before booking. Stage 02 before final invoice. Deliberately without proprietary tool dependence.

The workflow follows the responsibility separation. Stage 01 clarifies, before booking, whether the publisher meets the preconditions at domain level. Stage 02 checks, after publication and before payment, whether the publisher has honoured the briefing. Stage 01 requires only browser and command line. Stage 02 additionally requires the public Google Rich Results Test, reading competence and contract access.

The responsibility clarification is explicit. Stage 01 protects the publisher against retroactive invoice reduction, because domain policy sits outside the briefing frame. Stage 02 couples briefing compliance to invoice recognition.

Three-class assignment of the final invoice

Citation-Buy, Mixed-Buy, Mention-Buy. Plus two differentiated FAIL rows.

The combination of A-class fulfilment and B-class count yields a three-class assignment of the final invoice with two differentiated FAIL rows. The FAIL rows follow the responsibility separation from Section 02.

Mention-Buys retain residual value in the citation layer. Ahrefs documented in 2025 that brand mentions without a link in LLM answers produce brand-recall effects. The residual value is billed at the mention price, not at the citation price. This separation prevents the common market error of procuring a mention deliverable at the citation price and asserting a citation.

The final invoice follows a multiplicative formula: final_price = list_price × criteria_factor × model_blended_factor. The model-blended factor folds the principal’s engine reach into the price. The complete price-factor model sits in Codex chapter 3.

In-house reading or mandate reading?

The same eighteen criteria at the core. Different operational depth.

The eighteen criteria are formulated as a procurement standard for mandate work, but carry without a mandate relationship as an internal audit tool for teams with their own content offensive or their own advertorial procurement.

Both readings use the same eighteen criteria at the core, at different operational depth. The in-house reading is the entry level, the mandate reading is the full version with the contract lever.