How to Audit a Competitor's Entity Infrastructure in 30 Minutes

Most people audit competitors by looking at their website. Maybe their social media. Maybe their Google ranking for a few keywords. That tells you what they are saying. It tells you almost nothing about what machines believe about them.

There is a different kind of audit. One that maps the entity infrastructure a competitor has built (or failed to build) across the systems that AI search agents actually query. Knowledge Graph entries. Wikidata records. Schema markup. ORCID profiles. Verified cross-platform links. The structural layer underneath the content.

This audit takes 30 minutes. It requires zero paid tools. And when you are done, you will have a scored assessment of exactly where a competitor stands in machine-readable verification, and exactly where the gaps are that you can fill first.

I run this audit on competitors across all three of my companies. Industrial pumps, craft publishing, entity infrastructure consulting. The protocol is the same regardless of industry. Machines do not care about your vertical. They care about your verification.

Why entity audits matter now

Traditional competitive analysis asks: what keywords do they rank for? What content are they publishing? What backlinks have they earned?

Entity competitive analysis asks a different question: can AI agents verify who they are?

This matters because AI search does not rank pages. It selects entities to cite. ChatGPT, Gemini, Perplexity. They all pull from structured knowledge sources. If your competitor has a Wikidata entry and you do not, that is not a branding gap. That is a verification gap. It means machines can confirm their existence through an independent knowledge base and cannot confirm yours.

The companies winning in AI search are not the ones with the best content. They are the ones with the most complete trust chain, the layered verification that gives AI agents confidence to cite them by name.

An entity audit reveals the specific layers your competitor has completed. And, more importantly, the layers they have not.

Before you start: what you need

A browser. A spreadsheet (or a plain text file, I will not judge). The competitor's primary domain and the name of their key person (CEO, founder, or public-facing authority). That is it.

No paid tools. No API access. No crawlers. Everything in this protocol uses publicly available information, because that is exactly what AI agents use.

One rule: audit one competitor at a time. Do not try to batch five at once. Depth beats breadth here. You want to understand their complete entity picture, not skim five surfaces.

The 30-minute audit protocol

Seven steps. Each one maps to a specific layer of entity infrastructure. I have included estimated time for each step, but do not stress about the clock. Some steps take 2 minutes, some take 5. The point is that the whole thing fits in a single focused session.

Step 1: Google Knowledge Panel check (3 minutes)

Search Google for the competitor's company name and their key person's name. Two separate searches.

What you are looking for: a Knowledge Panel on the right side of search results. That rectangular box with structured information, logo, description, social links, "People also search for."

Score this:

• Full Knowledge Panel with verified info: They have strong entity recognition. Score 3.
• Partial panel (just basic info, no social links, sparse): Entity exists but is incomplete. Score 2.
• No panel at all: Google does not recognize them as a distinct entity. Score 0.

Do the same for their key person. A personal Knowledge Panel is harder to earn and signals strong individual entity authority.

As I covered in reading a competitor's Knowledge Panel, the panel itself is a map. Every field that is populated represents a verification layer they have completed. Every empty field is a gap.

Step 2: Wikidata and Wikipedia presence (3 minutes)

Go to wikidata.org. Search for the competitor's company name and their key person's name.

Wikidata is the structured database that feeds Wikipedia, Google Knowledge Graph, and most AI systems. If an entity exists here with properties filled in (founded date, headquarters, official website, social media links), machines treat it as a verified reference point.

What to record:

• Does a Wikidata item exist for the company? Note the Q-number (e.g., Q12345).
• Does a Wikidata item exist for the key person?
• How many properties are filled? (Fewer than 5 is sparse. More than 15 is robust.)
• Is there a Wikipedia article linked from the Wikidata entry?

Score: Wikidata with 10+ properties = 3. Wikidata exists but sparse = 2. No Wikidata entry = 0.

Step 3: Schema markup audit (5 minutes)

Visit the competitor's homepage. Right-click, View Source. Search for "application/ld+json" in the page source.

Alternatively, use Google's Rich Results Test (search.google.com/test/rich-results) and paste their URL. It will show you exactly what structured data Google can read.

What to check:

• Organization schema: Does it exist? Does it include sameAs links to LinkedIn, Wikidata, Crunchbase?
• Person schema: Is their key person declared with jobTitle, worksFor, sameAs?
• Service schema: Are their services declared as structured entities, not just page text?
• @id consistency: Do schema entries use consistent identifiers across pages, or are they disconnected fragments?
• Article/BlogPosting schema: Is their content marked up with author, datePublished, publisher?

Score: Comprehensive schema with sameAs and nesting = 3. Basic schema (just Organization name/logo) = 1. No schema or plugin-default only = 0.

Step 4: Cross-platform identity verification (5 minutes)

Check whether the competitor maintains verified profiles on the platforms AI agents cross-reference. Visit each one and search for the company or key person:

• LinkedIn: Company page + personal profile. Are they active? Is the information consistent with their website?
• ORCID (orcid.org): Search for the key person. ORCID is not just for academics. It is a persistent identifier that AI agents trust for individual verification.
• Google Scholar: Does the person or company have a profile? Any publications?
• Crunchbase: Company entry? Funding data? Team listed?
• Google Business Profile: Claimed and active? Reviews? Photos?
• Industry directories: Trade associations, professional registries relevant to their sector.

Count the platforms where they have a presence. Then check for something more important: consistency. Is the company name spelled the same way everywhere? Is the description aligned? Are the URLs pointing back to the same domain?

This is where most competitors fall apart. They have profiles everywhere, but the information does not match. Different company names, outdated descriptions, broken links. AI agents see that inconsistency and reduce their confidence score.

Score: 5+ platforms with consistent info = 3. 3-4 platforms or inconsistent info = 2. Fewer than 3 or major inconsistencies = 0.

Step 5: AI citation test (5 minutes)

This is the reality check. Open ChatGPT, Gemini, and Perplexity. Ask each one: "Tell me about [Competitor Name] and what they do." Then try: "Who is [Key Person Name]?"

What to observe:

• Does the AI know them at all?
• Is the information accurate and current?
• Does the AI mention their specific services, achievements, or publications?
• Does the AI confuse them with someone else? (Name collision is a real problem.)
• Does the AI cite them when you ask category questions? ("Who are the leading companies in [their industry]?")

This step connects directly to the AI visibility audit protocol. What you are doing for a competitor is the same thing you should have already done for yourself. If you have not, do that first.

Score: Recognized by all 3 AI platforms with accurate info = 3. Recognized by 1-2 with partial info = 2. Unknown or confused = 0.

Step 6: Content and authority signals (5 minutes)

Check the competitor's content for signals that build entity authority over time:

• Published works: Do they have books (check WorldCat, Google Play Books, Amazon)? Research papers? Industry reports?
• Speaking engagements: Conference talks listed on their site or on event pages?
• Media mentions: Search "[Competitor Name]" in Google News. Any press coverage?
• Third-party citations: Are other websites citing them as a source? Not just linking. Actually citing their work or data.
• rel=me links: Check their homepage source for rel="me" attributes connecting to their profiles. This is a verification signal machines use.

Score: Multiple published works + media mentions + third-party citations = 3. Some content authority but limited external validation = 1. No published works, no mentions, no citations = 0.

Step 7: Freshness and velocity (4 minutes)

Entity infrastructure is not a one-time build. Machines pay attention to freshness. Check:

• When was their website last updated? (Check blog dates, news sections, footer copyright year.)
• When was their Wikidata entry last edited? (Click "View history" on the Wikidata item.)
• Are they actively publishing? (Blog frequency, LinkedIn posting frequency, new publications.)
• Are their schema declarations current? (Do they reflect current services, current team, current contact info?)

A competitor who built solid entity infrastructure two years ago but has not maintained it is losing ground. AI agents weight recency. Stale information gets deprioritized.

Score: Active updates in the last 30 days across multiple channels = 3. Some activity but sporadic = 1. Dormant for 3+ months = 0.

The audit scorecard

Here is the full scorecard. Maximum score is 21 points (7 checkpoints, max 3 each). I have added the specific things to look for and what each score level means.

Interpreting the total:

• 0-7: Weak entity infrastructure. Major gaps across most layers. You can overtake them on verification alone without touching their content game.
• 8-14: Partial infrastructure. They have built some layers but have obvious holes. Targeted action on their weak checkpoints gives you an advantage.
• 15-21: Strong infrastructure. They take entity verification seriously. Competing requires matching their foundation and then differentiating on depth, freshness, or niche authority.

The audit process flow

What to do with the results

The scorecard is not a trophy. It is a map.

Every checkpoint where a competitor scores 0 is a layer of entity infrastructure you can build before they do. And in entity verification, first-mover advantage is real. The entity that Wikidata recognizes first becomes the canonical reference. The Knowledge Panel that exists first sets the baseline that competitors have to differentiate against.

Here is how I prioritize based on audit results:

If they score 0 on Wikidata but you also score 0: Race to Wikidata. Create your entry first, with real citations. Make yourself the established entity in the knowledge base before they show up.

If their schema is basic and yours is too: Implement comprehensive schema with sameAs links, nested relationships, and service declarations. This is the fastest win because it only requires changes to your own website.

If they score 0 on AI citation: That means machines do not trust them enough to cite. If you build your trust chain while they are absent, AI agents will learn to associate your name with the category queries that matter.

If they score 3 on everything: Respect that. Then look at what they are NOT doing. Maybe their entity infrastructure is strong for their company but weak for their individual leaders. Maybe their services are well-declared but their geographic specificity is missing. There is always a dimension they have not covered.

Common patterns I see

After running this audit dozens of times across different industries, patterns emerge.

Pattern 1: Great content, zero infrastructure. They blog regularly. They post on LinkedIn. They might even rank well on Google. But they have no Wikidata entry, no structured schema, no ORCID, no cross-platform consistency. Their content exists. Their entity does not. This is the most common pattern and the easiest to exploit.

Pattern 2: Technical foundation, no velocity. Someone set up their schema and Wikidata entry two years ago. It was good work. But nothing has been updated since. Their founding date is right but their current services are not listed. Their team page has people who left. Machines notice when declared information conflicts with recent evidence.

Pattern 3: Platform presence, no connection. They have LinkedIn. They have a Google Business Profile. They might have Crunchbase. But nothing links back to anything. No sameAs in their schema. No rel=me on their website. Each profile is an island. AI agents see five separate fragments instead of one verified entity.

Pattern 4: Person strong, company weak. The founder has great personal entity authority. Published author, speaking circuit, maybe even a Knowledge Panel. But the company entity is barely established. This is common in consulting and professional services. The problem: if the person leaves or steps back, the company's entity infrastructure collapses.

Running the audit on yourself first

I should say this clearly: do not audit a competitor before you audit yourself.

Run the exact same 7-step protocol on your own company and your own name. Score yourself honestly. If you score 8 and your competitor scores 14, the audit just told you something important. Not about them. About you.

The audit is most valuable when you have both scores side by side. Your score shows where you are. Their score shows what is possible. The gap between them shows what to build next.

This is not complicated. It is just work that most people skip because they are busy producing content instead of building the infrastructure that makes content findable by machines.

What this does not tell you

This audit maps entity infrastructure. It does not measure content quality, customer satisfaction, actual expertise, or business viability. A competitor can score 21 on this audit and still deliver terrible work. A brilliant practitioner can score 3 because they never bothered with verification.

The audit measures machine readability, not human value. Keep that distinction clear.

What it does tell you: in the specific arena of AI search and entity verification, here is where they stand and here is where you can move. That is a narrow but increasingly important slice of competitive positioning.