Last updated: August 14, 2026 The Methodology

The Algorithmic Authority Stack™

Seven structural layers decide whether AI systems resolve you, classify you, and name you when a buyer asks for a shortlist. Each layer has one failure mode. The diagnostic tests all seven and finds the one breaking first.

73–86% of tested vendors get zero AI citations, across four industries

20 + 480 companies scored in Wave 1, vendors measured across Wave 2

64–71% of names returned sit in a category next to the one asked about

Source: The Algorithmic Authority Index, Waves 1–2, 2026

Find out which layer is breaking Read the research

What this measures

AI parses structure. It does not read your about page.

AI systems evaluate identity consistency, semantic density, citation patterns, and trust markers. None of that is how Google ranking works, which is why AI visibility is a separate discipline from SEO. The data on that split is in Google rankings and AI citations.

Every query where a competitor gets named and you don't is a recommendation you'll never know you lost. AI answers have no page two. You are either in the answer or absent from the conversation. That gap between human visibility and machine visibility is The Great Decoupling.

The Stack is sequential. Each layer depends on the one below it. Layer 4 fixes fail while Layer 1 is broken. Most companies start with content, and that is why most companies stay invisible. Before Layer 1, one prerequisite: AI crawlers have to reach your site at all. Fix AI crawlability first if you are unsure.

The framework came out of pattern recognition across 50+ B2B audits, then got tested against original research. Wave 1 scored 20 companies across 420 assessments. Wave 2 measured four industries at category scale. Everything on this page traces to the published Index.

The seven layers

Each layer has one failure mode. Find yours.

Market Identity Clarity

Failure mode: Identity Fragmentation

Your LinkedIn says "AI solutions provider." Your site says "intelligent automation platform." Your guest bio says "technology consultant." AI registers three companies and disqualifies all three.

What the diagnostic tests: cross-surface identity consistency across LinkedIn, website, directories, press, and bios. How many descriptions exist. Whether AI resolves you to one entity.

From the research: in Wave 1, 16 of 20 B2B companies carried conflicting identity signals across five or more surfaces. In Wave 2, the old brand name still returned in 78% to 81% of rebrand queries, years after the change.

Fix guide· Run the test· Why AI describes you wrong· After a rebrand· Entity resolution

Expertise Architecture

Failure mode: Authority Collapse

You have 20 years of expertise and AI can verify none of it. The knowledge lives in your head, your meetings, and your experience, where no machine can reach it.

What the diagnostic tests: published expertise signals, citation by authoritative sources, co-occurrence with recognised entities, and credential verification pathways.

From the research: Wave 1 found revenue predicts almost nothing about authority framing. A challenger with narrow category ownership scored 84 while a public company with ten times the revenue scored 83, and only one of them got a category-ownership claim from the model.

Fix guide· Why AI does not pick the best expert· The founder track

Semantic Density

Failure mode: Semantic Drift

You call it "workflow automation" on one page, "process optimization" on another, and "business efficiency tools" on a third. To a human those are synonyms. To AI they are three topics, and your authority splits three ways.

What the diagnostic tests: term consistency across surfaces, drift measurement, whether core concepts carry one term or twelve, and how that compares against competitors.

From the research: Wave 2 found 64% to 71% of returned vendors sat in a category adjacent to the one the buyer asked about. The buyer receives a confident shortlist built on the wrong category and cannot see the error.

Fix guide· The Semantic Drift Epidemic

Training-Ready Content

Failure mode: Citation Invisibility

You publish twelve posts a month and AI cites none of them. Readable content and extractable content are different things, and only one of them gets quoted.

What the diagnostic tests: data density per article, source citation patterns, expert attribution signals, and whether your content is structured for machine ingestion or just filling a blog.

From the research: this layer failed universally in both waves. Salesforce earned zero citations on the AI CRM question. Deloitte earned zero on digital transformation. In Wave 2, 100% of tested fintech vendors failed training-ready content, and so did 100% of emerging healthcare startups. It is the single biggest fixable gap in B2B.

Fix guide· The 43,000:1 problem· Does your site pass the machine test?· Non-commodity content· Writing with AI

Algorithmic Touchpoint Presence

Failure mode: Surface Dependency

Visibility on one platform carries almost nothing to the others. Each system builds its own answer set from its own source pool.

What the diagnostic tests: presence across AI-indexed surfaces, platform-specific citation patterns, which sources each system actually pulls from, and the gap between where you are and where citations happen.

From the research: Wave 2 documented Trust Seed Substitution. When brand content fails extraction, AI routes evidence elsewhere. Clinical queries went to peer review and FDA databases. Fintech queries went to FinCEN, the CFPB, and trade press. Administrative queries went to G2 and Reddit. Most vendors invest on none of the surfaces that decide their shortlist.

Read Layer 5· Getting cited by Perplexity· Reddit for B2B visibility· Why AI ignores your videos

Trust & Proof Signals

Failure mode: Trust Gap

AI evaluates trust before quoting anyone. Without third-party corroboration, even excellent content gets passed over for a source with stronger markers.

What the diagnostic tests: third-party mention frequency, backlink quality and authority, press coverage, industry recognition, and verifiable credentials against competitor trust profiles.

From the research: healthcare returned the strictest trust behaviour in the study. 86% category absence, and 42% of clinical queries refused outright. Each vertical sets its own evidence bar, and Wave 2 documents where each one sits.

Fix guide· E-E-A-T for AI models

Visibility Measurement

Failure mode: Measurement Blindness

You cannot optimise what you are not tracking. AI citations are stochastic, so the same query returns different answers at different times, and a single screenshot proves nothing.

What the diagnostic tests: your current citation baseline, brand mention accuracy in generated answers, citation share against competitors, and a measurement framework you can run on a cadence.

From the research: the Index measures Share of Model, the percentage of AI responses in a query set that name a specific vendor. Above 5% composite is a Category Winner. Below 1% is Category Absence. Across four industries, 73% to 86% of tested vendors sat below that line, and almost none of them knew.

Fix guide· What is Share of Model?· When AI stops recommending and starts selecting

The Algorithmic Authority Index™

Three companies that should be impossible to miss. All three were missed.

Category leadership, publishing volume, and brand recognition each failed to produce citations. The structure decided the outcome in every case.

Salesforce · Layer 4

Zero citations of salesforce.com when Perplexity is asked about the best AI CRM approach.

A $41B category creator, absent from its own category question.

Deloitte · Layer 4

Zero results cite deloitte.com on digital transformation queries.

Thousands of thought leadership pieces a year. Cited on none of them.

BILL · Layer 1

Bill.com still wins 91% of queries after the company became BILL.

The rebrand reached humans. The models kept the old name.

Wave 1, March 2026: 20 companies, 420 assessments. Wave 2, July 2026: four industries. The industrial, healthcare, and fintech deep-dives ran 900 documented sessions each. Cybersecurity used a separate design against a universe of roughly 3,000 active vendors. Tested across ChatGPT, Perplexity, and Google AI Overviews. Read the full Index.

The pattern library

Four diagnostics. Four different broken chains.

Anonymised companies across four industries. Each one found a different layer breaking first, which is the whole argument for testing all seven before fixing any.

Industrial B2B The category leader losing to a new entrant Layer 1 · Identity Fragmentation SaaS hybrid A 47-point citation gap between the B2B and B2C sides of one company Layer 1 · Entity foundation Consumer goods The Polyglot Penalty, and what it costs across markets Layer 1 · Identity Fragmentation Regulated retail An editorial authority gap in a category with a high evidence bar Layer 4 · Training-Ready Content

View the full pattern library, or read why your company is invisible to AI and what AI actually sees when it looks at you.

Where you take this next

Same methodology. Two entities it can run on.

The Stack tests the same seven layers whether the entity is a company or a person. The ladder is identical: verify, diagnose, then build.

Start with the free assessment. Six questions, about 90 seconds, and it names the layer breaking first. To verify that against live AI answers, the Snapshot is a 48-hour written diagnostic at $497, and the fee credits in full toward the Audit within 30 days.

Track 1 · Brand authority

A B2B company that needs AI to recognise the corporation.

Snapshot $497
Audit $4,500
Implementation $4,000 per month

See the Audit →

Track 2 · Founder visibility

A founder, fractional executive, or agency principal who needs AI to recognise them personally.

Snapshot $497
Audit $4,500
Implementation $4,000 per month

See the founder track →

Audit plus the first three months of Implementation runs $16,500. Not sure which track fits, or want to understand the role before you buy anything? Here is what an AI Visibility Architect actually does.

Find out which layer is breaking Or run the $497 Snapshot

Go deeper

Three places to start.

The definitions What Is Answer Engine Optimization (AEO)? The symptom Why Your Competitors Show Up in ChatGPT and You Don't The fixes Every fix guide, one per layer, free and complete