Table of Contents

Proven Entity Engineering Delivers 70% ROI

A $50M-$150M industrial manufacturer. Four starting conditions. Four documented outcomes.

---

He opened his laptop on a Tuesday morning. It was an ordinary moment, the kind that often comes before an extraordinary realization. The marketing director at a $100M precision manufacturer had run the company’s commercial presence for eleven years.

He knew the trade show calendar. He knew which publications his buyers read. He could name the three competitors his sales team mentioned most. But he couldn’t name—because he’d never seen it measured—the category of loss that was about to become visible.

He typed a query into ChatGPT. Not his company’s name. He asked the question a serious buyer would ask at the start of an evaluation. That question was: who makes what we make, at our quality, in our markets?

Three companies came back. They were described with confidence, attributed independently, and named without hesitation. His company wasn’t one of them. This article explores how entity engineering addresses such AI invisibility.

He ran the same query on Perplexity. Then on Google AI Overviews. The results were consistent. His competitors appeared. His company did not. Eleven years of work. A well-maintained website. An active content program. A sales team that closed real business.

None of it had built the one thing the AI systems now required: a machine-readable identity record that would get his company cited, not ignored. The machine didn’t know they existed. And because of that, neither did the buyers who were deciding whether to call.

---

What Is Your Entity Engineering Starting Position?

The company I’ll describe next started from the exact same position. Not similar—identical. Four conditions, each documented.

1. First: invisible in AI-generated answers. When their buyers ran category-level queries on AI platforms, this company didn’t appear. Not in a qualified way, not as a secondary mention. They were simply absent. The machine had nothing to anchor to, so it left them out.
2. Second: no Knowledge Panel presence. That structured entity card—the one that confirms a company exists and tells what it does—didn’t exist for them. The foundation layer of AI recognition had never been built.
3. Third: zero AI citation across the three platforms their buyers used. ChatGPT, Perplexity, Google AI Overviews. Their buyers were using these to build shortlists and validate options. This company had zero citations across all three.
4. Fourth: competitors appeared where they should have. This is what made the first three conditions concrete. The absence wasn’t a general AI problem affecting the whole category. It was specific. Competitors were named with confidence. This company wasn’t. The gap was structural, and it had commercial consequences that never showed up in their CRM or reports.

The results of these four conditions are detailed in Articles 2, 3, and 4 of this series: pre-pipeline eliminations, late-stage confidence collapses, evaluations that ended without this company’s name. The root cause was always the same: AI visibility was zero. Not low. Zero.

The entity infrastructure needed for AI recognition didn’t exist. Their entire marketing investment had missed it because it wasn’t designed to address it. In B2B buying, 94% of buyers now use AI for vendor research.

Sixty percent of the buyer’s journey is over before they ever contact a vendor. The shortlists they bring into that first conversation are built largely from AI responses. A company absent from those responses is absent from those shortlists. And a company absent from the shortlist has about a 5% chance of winning the deal. This company was absent from every shortlist built by every buyer who used AI.

That cost never appeared in their data, because the loss happened before the data was created. Every buyer who consulted AI before making first contact saw competitors. They never saw this company. This article provides the methodology to fix that.

How Our 4-Layer Entity Engineering Build Works

What changed wasn’t their product, pricing, sales team, website, or trade shows. What changed was the infrastructure layer that AI systems consult when answering questions about who belongs in a category. The intervention was an entity engineering build executed through the AI Authority Method—our implementation of the Entity Engineering discipline. The build followed all four layers of the architecture in sequence, with verification required at each gate before moving forward.

1. Layer 1: Delivery and Render. Before any entity records could be registered, AI platforms needed to reach and parse the company’s content without obstruction. We audited and corrected structured data, crawlability, and render architecture against our spec. This layer is invisible—no new rankings or citations. But without it, nothing else works. The wiring had to be confirmed first.
2. Layer 2: Entity Identity. This is the knowledge graph layer. We established a canonical, machine-readable entity record in authoritative registries AI systems use: authoritative generalized and industry databases and verified profiles. An interconnected network of third-party sources consistently describing what this company was and what category it belonged to. This layer produces Knowledge Panel presence. Its absence had made the company invisible.
3. Layer 3: Citation Engineering. The corroboration network. We conducted structured outreach to authoritative third-party sources so the claims the company made about itself were independently confirmed. Not press releases—systematic corroboration. This builds the evidence base that makes AI cite an entity with confidence, not hesitation.
4. Layer 4: Narrative Engineering. The framing layer. We established the terminology, category signals, and competitive context that guide how AI describes the company. By the end of this layer, AI would describe them in the language they’d choose, in the category they’d earned, with the positioning they’d built over eleven years of excellent work—work the machine previously couldn’t attribute correctly.

The infrastructure these four layers produce is fundamentally different from standard marketing output. SEO produces rented positions. Content produces rented citations. PR produces rented credibility.

Entity engineering produces owned infrastructure—records in authoritative systems that the company controls, that persist without ongoing spend, and that compound in AI training data over time. We were establishing algorithmic authority: the compound recognition signal that builds as AI increasingly associates an entity with its category.

The Authority Build phase took twelve weeks, with gate verification at each layer. The subsequent Compound maintenance period let recognition signals propagate and compound across AI training cycles. The full engagement—Build plus maintenance—lasted eighteen months. Each layer addressed one of the four starting conditions. Each gate confirmed that condition was closed before the next layer began.

What Results Does 18 Months of Entity Engineering Deliver?

Months after the build started, something had changed in every AI response relevant to this company’s buyers. Buyers opening ChatGPT and typing a category-level query—the kind a Director of Operations asks when building a supplier list—now found this company. Named. Described confidently. Cited by independent sources the AI trusts. Not hedged with “reportedly” or “claims to be.” Present, attributed, and authoritative. The three documented outcomes:

1. $6M in revenue attributable to AI discovery. The mechanism was clear: buyers arrived at conversations already informed. The AI had told them who this company was and why they were worth considering—before the sales team got the call. These buyers didn’t find the company through search and then use AI. Their research was done entirely in AI systems. Their shortlists came from AI responses. They contacted this company because AI placed it in the category conversation. The revenue is documented. The attribution to AI-driven discovery is documented.
2. 70% ROI. This is the ratio of the $6M documented revenue impact to the investment in the entity engineering engagement. Let’s be precise: this isn’t a modeled projection. It’s the actual ratio of outcome to cost for this single engagement, in this sector, over 18 months. The $6M happened. The engagement cost what it cost. The math is the math.
3. 98% knowledge graph coverage across all target AI platforms. The structured entity card that confirms a company’s presence was active and accurate where their buyers looked. 98% of targeted platform queries returned a confirmed knowledge graph node.

Metric	Starting Position	18-Month Result
Knowledge graph coverage	None	98% across all target AI platforms
AI Citation — Target Platforms	Zero	Confirmed across ChatGPT, Perplexity, Google AI Overviews
Revenue Attributable to AI Discovery	$0 documented	$1M+
ROI on Engagement	N/A	70%+

These are not projected figures. They are documented outcomes from an Entity Engineering engagement. The methodology produced them. The methodology is replicable. This article proves it.

What These Entity Engineering Results Mean for Your Company

We didn’t pick this company because entity engineering worked unusually well for them. We picked them because they’re the profile: $50M–$150M Capital Goods manufacturer, precision components, strong in-person reputation built over decades, solid marketing stack, active sales team—and zero AI recognition.

Not thin. Not inconsistent. Absent.

This company has the same revenue profile as yours. The same buyer base. The same AI invisibility problem.

This is the first fully documented engagement at this methodology scope. What we have now is proof of concept: evidence that the methodology works at this scale, in this sector, with this buyer profile.

If you’re a manufacturing CEO reading this and thinking that company sounds like ours, you’re right. The profile is the profile. The starting position is the starting position. The methodology that closed the gap is systematic, documented, and available.

It’s not a guarantee of a specific outcome. It’s evidence that the outcome is achievable.

The entity infrastructure built for this company is still operating. No campaign is running. The knowledge graph records exist. The Knowledge Panel is active. The corroboration network is in place.

New buyers who consult AI today find the same company that was invisible 18 months ago. That’s what “permanent” means for entity infrastructure: the investment was made up front, and the AI visibility persists.

Go back to the opening scene. The marketing director. Tuesday morning. ChatGPT open. Category query typed. Three competitors named. His company absent. That diagnostic is available to you right now. Open an AI platform—ChatGPT, Perplexity, Google AI Overviews. Type the category query your best buyers use when they start evaluating suppliers. Not your company name.

The question they ask when they don’t yet know who to call. See what appears. See what doesn’t. The gap between what appears and what should appear is the gap entity engineering closes. The diagnostic is the same one they ran. The question is whether your result will be different. This article provides the framework to ensure it is.