Part 3 — The OMG Protocol structure
3.1 The three pillars
The OMG Protocol organizes authority engineering into three pillars: Optimize, Manifest, Generative. The order is not arbitrary. It corresponds to the three minimum gates a brand passes through to be cited in an AI-generated answer.
| Pillar | Gate | Failure mode |
|---|---|---|
| Optimize | Can AI systems engage with the brand at all? | The brand cannot be crawled, parsed, or structurally understood by automated systems |
| Manifest | Does the engagement produce an accurate representation? | AI systems form an inaccurate or shallow understanding of what the brand is |
| Generative | Is the represented brand trusted enough to be cited? | The brand is well-represented but not surfaced because external validation is absent |
Failure at any pillar is sufficient to prevent citation. A brand may be technically pristine (Optimize complete) and content-rich (Manifest complete) yet receive no AI citations because external authority signals are absent (Generative deficient). Conversely, a brand may have abundant external authority (Generative strong) but receive degraded citations because AI systems cannot parse its content (Manifest weak). The three pillars are independently necessary.
The pillars also map to distinct organizational ownership and distinct rhythms of work:
| Pillar | Owning team archetype | Rhythm of work |
|---|---|---|
| Optimize | Web engineering, technical SEO | Sprint-shaped — work decomposes into discrete fixes that ship and verify quickly |
| Manifest | Content, editorial, subject-matter experts | Cycle-shaped — work decomposes into editorial commitments that span content production and ingestion windows |
| Generative | Communications, public relations, partnerships, knowledge graph operations | Campaign-shaped — work spans relationships, third-party publication, and authority compounding |
A single team attempting to address all three pillars simultaneously typically fails at all three. Optimize requires engineering velocity. Manifest requires editorial discipline and domain expertise. Generative requires relationships and time. Conflating them in operational planning is the most common reason AVO programs stall — the team scopes work without distinguishing the rhythms, and the slowest rhythm becomes the bottleneck for all of them.
For the practitioner scoping engagement structure, the three-pillar mapping informs which roles within the brand’s organization (or which external partners) are involved in which work. A brand without engineering velocity will struggle on Optimize. A brand without editorial capacity will struggle on Manifest. A brand without communications relationships will struggle on Generative. The practitioner’s role in scoping includes identifying these capacity questions before the work begins, not discovering them mid-engagement.
3.2 Why three pillars and not two or four
The three-pillar structure is not arbitrary count. It maps to the three minimum gates a brand passes through to be cited in an AI answer. A discipline with fewer pillars conflates gates that should be measured separately. A discipline with more pillars fragments interventions across team boundaries that do not exist in real organizations.
Two-pillar alternatives typically merge Manifest and Generative into a single “content authority” pillar. This conflation is an error because it obscures whether a brand’s invisibility is a content problem (the brand has not produced sufficient depth or quality of content) or a recognition problem (the brand has produced excellent content but no third party has validated it). These are different problems with different solutions, and conflating them causes practitioners to apply content-pillar interventions when recognition-pillar interventions are needed, or vice versa.
Four-pillar alternatives typically split one of the existing pillars in two. For example, separating “structured data” from “technical health” into distinct pillars. The split is unproductive because the actions within Optimize all share the same operational rhythm (sprint-shaped, engineering-owned) and the same measurement scaffolding. Splitting them into separate pillars produces administrative overhead without diagnostic benefit.
The three-pillar count is the minimum that preserves the diagnostic distinction between consumption (can the AI use the site at all), representation (does the AI form an accurate picture), and trust (does the AI rely on the picture). These are the three distinct failure modes that produce non-citation, and they are the three distinct interventions an AVO program must be able to direct independently.
3.3 The pillar–team–horizon mapping
Beyond the abstract pillar definitions, the practical implementation of each pillar requires distinct organizational capacity. This mapping is the most common point of friction in early-stage engagements, and the practitioner should make it explicit during engagement scoping.
Optimize
Owning team archetype: Web engineering, technical SEO, devops.
What this team must be capable of: shipping technical changes to the brand’s web properties, managing infrastructure (CDN, server configuration, performance), implementing structured data (Schema.org markup), maintaining canonical URL structure, configuring crawler permissions (robots.txt, robots meta tags).
Bottleneck signals: a brand with no in-house engineering capacity (or whose engineering capacity is fully occupied with other priorities) will be Optimize-bottlenecked regardless of AS finding. The practitioner should identify this bottleneck during scoping. Solutions include external engineering retainers, engagement-scoped engineering hires, or scope reduction to engagement-feasible Optimize work.
Manifest
Owning team archetype: Content, editorial, subject-matter experts.
What this team must be capable of: producing original long-form content with editorial standards, citing sources accurately, maintaining publication and update cadences, organizing content into pillar-and-cluster structures, refreshing existing content as facts change.
Bottleneck signals: a brand producing thin or duplicative content, or relying entirely on syndicated or aggregated content, will be Manifest-bottlenecked regardless of AS finding. The practitioner should identify this bottleneck during scoping. Solutions include hiring editorial staff, engaging external content production with editorial oversight, or commissioning subject-matter experts for specific content programs.
Generative
Owning team archetype: Communications, public relations, partnerships, knowledge graph operations.
What this team must be capable of: building and maintaining media relationships, commissioning original research, executing Wikipedia and Wikidata work with policy compliance, earning citations from authoritative publications, managing the brand’s structured-knowledge presence across multiple platforms.
Bottleneck signals: a brand with no communications function, no media relationships, and no original research output will be Generative-bottlenecked regardless of AS finding. The practitioner should identify this bottleneck during scoping. This is the most expensive bottleneck to remedy and the most often misjudged at engagement start. A brand stakeholder who states “we’ll handle the communications work in-house” without a dedicated communications function will produce minimal Generative-pillar progress.
The work of engagement scoping includes mapping the brand’s actual organizational capacity to these three teams’ archetypes and identifying which gaps need to be addressed before AVO loop work can produce results. A brand may need to hire, contract, or scope-reduce before the loop can run effectively.
3.4 The six measurement vectors
Within the three pillars, OMG specifies six vectors. Each vector groups datapoints that share a measurement type and an intervention type. Vectors are the level at which strategic prioritization happens — the question what should we work on next is answered at the vector level, before drilling down to specific datapoints.
| Pillar | Vector | What it covers |
|---|---|---|
| Optimize | V1.1 Signal Architecture | Structured data, schema markup, semantic HTML, metadata quality |
| Optimize | V1.2 Technical Health | Crawler accessibility, performance, security, canonical consistency, multilingual setup |
| Manifest | V2.1 Semantic Density | Topical relevance, content depth, attribution, entity recognition, claim density, originality, structural quality |
| Manifest | V2.2 Structural Legibility | Hierarchy, formatting, accessibility, identity clarity, update signals, chunk extractability |
| Generative | V3.1 Knowledge Validation | Citation strength, AI citation presence, Wikidata presence, knowledge graph depth |
| Generative | V3.2 Trust Alignment | Domain authority, trust signals, transparency indicators, external validation, content freshness |
The within-pillar splits are diagnostic distinctions:
| Split | What it isolates |
|---|---|
| V1.1 / V1.2 within Optimize | Signals (machine-readable structure) versus infrastructure (the site’s ability to be crawled and rendered). Both fail modes look like “AI cannot use this site,” but they require different fixes — V1.1 is content engineering, V1.2 is operational engineering. |
| V2.1 / V2.2 within Manifest | Substance (content depth, attribution, originality) versus legibility (structural clarity, chunkability, formatting). Content can be deep but unparseable, or legible but shallow. |
| V3.1 / V3.2 within Generative | Entity recognition (the brand exists in structured-knowledge systems) versus source trust (the brand is treated as authoritative by AI systems). A brand can be recognized but not trusted, or trusted but not recognized as the right kind of entity. |
For the practitioner, vector-level analysis is the primary level for action selection. A brand with low Optimize-pillar score and the deficit concentrated in V1.1 needs structured-data work; a brand with the same low Optimize pillar but the deficit concentrated in V1.2 needs infrastructure work. The two needs are addressed by different actions and different teams.
3.5 The thirty actions catalog overview
The OMG Protocol includes a catalog of thirty canonical actions — the operational work practitioners perform within the three pillars. The full action catalog is the subject of Document 2: The OMG Action Playbook. This Foundations document presents the catalog at the level required to understand the protocol’s structure.
The pillar distribution is asymmetric:
| Pillar | Action count | Examples |
|---|---|---|
| Optimize | 7 actions | Competitive analysis, analytics infrastructure, technical foundation, structured data foundation |
| Manifest | 10 actions | Question-based opportunity mapping, answer-first architecture, FAQ hubs, evidence-based content, content refresh |
| Generative | 13 actions | Entity verification, advanced topic clustering, long-form content, media outreach, original research, academic citations, Wikipedia/Wikidata, partnerships |
The asymmetry reflects operational reality. Optimize work is concentrated in a smaller number of high-leverage actions because the pillar’s scope (technical readiness) is narrower. Manifest and Generative work is distributed across more actions because each addresses a narrower aspect of a broader remit (content quality and external authority, respectively).
A complete OMG practice does not require all thirty actions in every engagement. A typical engagement covers a subset selected based on what AS measurement reveals: brand maturity, current pillar/vector imbalance, and operational priority. Action selection is part of the practitioner’s craft, directed by AS findings rather than by intuition.
The thirty actions are catalogued with action IDs (O-1 through O-7, M-1 through M-10, G-1 through G-13). The IDs are stable across the methodology and the bible — when the team refers to “G-11” in conversation, they refer to a specific named action with a defined chapter in the Action Playbook.
3.6 Action sequencing logic
The thirty actions are not selected at random or by intuition. They have prerequisite structure that the practitioner must respect.
Hard prerequisites are conditions that must be true before an action can be attempted at all. Attempting an action without its hard prerequisites does not produce reduced effect; it produces failure or actively negative outcome. Examples:
- G-11 (Wikipedia and Wikidata Optimization) cannot be attempted without G-1 (Entity Verification) substantially complete and at least 3-5 substantive third-party citations available. Attempting G-11 without these prerequisites produces rejected articles, deleted Wikidata entries, and persistent damage with the volunteer editor communities that govern these platforms.
- G-3 (Comprehensive Long-Form Content) cannot be attempted without G-2 (Advanced Topic Clustering) at least partially complete. Without topic clustering, long-form content is produced without the structural authority architecture that makes it citable.
- M-2 (Answer-First Content Architecture) cannot be attempted without M-1 (Question-Based Opportunity Mapping). Without M-1, the answer-first work is directed at questions the brand’s audience does not actually ask.
Soft prerequisites are conditions that improve an action’s success rate but are not strictly required. Examples:
- G-4 (High-Authority Media Outreach) is more successful when G-3 (Comprehensive Long-Form Content) is complete because journalists more readily cite brands with substantial reference content than brands without.
- G-9 (Academic & Niche Citations) benefits from G-3 and G-8 (Original Research) being complete because academic citation is more accessible to brands with publishable research output.
Stage assessment is the practitioner’s judgment of what stage of brand maturity an action belongs to. AVO recognizes implicit stages of maturity that govern which actions are appropriate. These stages are not time-bounded; they are state-bounded. A brand transitions from one stage to the next when AS measurement indicates the prior stage’s work is sufficiently complete.
The stages, named for what the brand is engineered for at each:
| Stage | What is being built | Pillar emphasis | What AS looks like |
|---|---|---|---|
| Foundations | Machine-readability and basic citability | Optimize-heavy with early Manifest | AS rising from floor through low-Developing range |
| Depth | Content authority and structural legibility | Manifest-heavy with continued Optimize | AS in mid-to-high Developing range |
| Authority | External validation and knowledge-graph presence | Generative-heavy with continued Manifest | AS approaching Strong band |
| Sustained authority | Compounding and maintenance | Balanced across all three with maintenance discipline | AS in Strong-to-Elite range |
Action selection respects stage. Foundations-stage brands should not attempt G-pillar actions; the prerequisites are not in place. Authority-stage brands should not abandon Optimize and Manifest work; the foundations require maintenance.
The Action Playbook (Document 2) makes the prerequisite structure and stage assessment explicit per action. Foundations introduces the structure so the practitioner reading the Playbook understands what the prerequisites mean.
3.7 The Acceleration Layer
Some organizations choose to accelerate AVO outcomes through paid amplification — paid social, paid search, paid distribution of authority-building content. This work is not part of OMG methodology because OMG measures and improves earned authority and earned visibility. Paid amplification is a different discipline operating on a different theory of change.
AVO recognizes the Acceleration Layer as an optional service category outside the thirty OMG actions. For the practitioner, the distinction matters for two reasons:
- Measurement integrity. AS and VS measure earned authority and earned visibility. Paid amplification produces visibility that is not earned and should not be conflated with VS measurement. A brand whose VS lifts because of paid distribution is not the same as a brand whose VS lifts because of authority compounding.
- Strategic clarity. Paid amplification is appropriate when the brand needs visibility now while authority builds. It is not a substitute for authority work. A brand that runs paid amplification continuously without authority work produces visibility that disappears the moment paid spend stops. AVO produces visibility that compounds and persists.
The practitioner’s framing of the Acceleration Layer to brand stakeholders is straightforward: paid amplification is sometimes the right tool for the brand’s commercial needs, but it is a separate discipline with separate measurement. AVO can recommend when acceleration is appropriate (typically during the gap between OMG action and visibility manifestation), but the work itself is outside AVO scope.