In part one of this series breaking down our Althea Score and brand visibility audit, we looked at the perception gap through Front-End SEO. Identifying a perception problem is only the first step. To fix it, you have to look at the technical infrastructure that allows LLMs to discover your business. Even with the perfect narrative, your technical setup might be hiding your expertise from the models that need to cite it.
At Althea Labs, we see it daily. Brilliant B2B brands suffering from technical invisibility. They have the expertise, but their outdated or insufficient infrastructure makes it easy for LLMs to ignore them.
Why Your Infrastructure is Part of the Score
To understand why your technical setup is a primary variable in the Althea Score, you have to acknowledge the fundamental shift in how B2B decisions are made. Modern marketing no longer serves a single audience. It serves two evaluators running in parallel.
- The Human Buyer: They seek an emotional and intuitive experience. They care about narrative, visual authority, and brand resonance. Humans want to know if they can trust you.
- The LLM: They seek an analytical and fixed experience. Machines care about data density, explicit relationships, and crawlable evidence. They want to know if they should cite you, and they look at your technical presentation to determine that.
Your technical infrastructure is the bridge between these two worlds. If your backend is messy, outdated, or blocked, the LLM half of your audience is effectively blind. You can win the human's heart with a beautiful story, but if the LLM cannot parse the facts to verify that story, it has nothing to cite.
Think of it this way: traditional SEO is about a book being on the right shelf so a human can find it. AI search is the model pulling the book off the shelf and reading it at the moment someone asks a question, then quoting from it in the answer. Training, where a model absorbs a huge corpus months before it is released, is a separate and much slower path. If your technical infrastructure is broken, the book is essentially locked. The AI knows it exists but cannot read what is inside.
The Three Pillars of Machine-Readable Infrastructure
During an Althea Score audit, we evaluate three specific technical pillars to determine whether an LLM treats your site as a primary source or overlooks it entirely.
1. Beyond the Robots.txt
The primary point of failure for many B2B brands is the configuration of the robots.txt file. Historically, this file functioned as a set of instructions for search engines to ignore specific directories. Today it also decides which AI crawlers can read your site, and which cannot.
While most B2B sites are optimized for traditional search crawlers, they often fail to account for the new generation of agents, and those agents do different jobs. OpenAI alone runs several: GPTBot crawls content that may be used to train OpenAI's models, OAI-SearchBot surfaces websites in ChatGPT's search results, and ChatGPT-User fetches a page when a user asks about it. Anthropic splits the same way: ClaudeBot for training, Claude-SearchBot and Claude-User for search indexing and live questions. CCBot is different again. It belongs to Common Crawl, a non-profit that maintains an open archive of web crawl data, and that archive is widely used to train models.
- The Reality: Blocking is common and uneven. Ahrefs found GPTBot was the most blocked AI bot, blocked by 5.89% of the roughly 140 million websites it studied (May 2025). Among the most widely used news websites in ten countries, the Reuters Institute found that 48% were blocking OpenAI's crawler by the end of 2023, and 24% were blocking Google's. These restrictions are typically a reactive measure intended to protect proprietary data or to prevent scraping.
- Why It Matters: Training and search are two different doors. Blocking GPTBot keeps your pages out of OpenAI's training data. It does not, by itself, keep you out of ChatGPT's answers. OAI-SearchBot is what lets ChatGPT search find, show and cite your pages, and OpenAI's crawler documentation recommends allowing it in robots.txt to help your site appear in ChatGPT search results. Anthropic says the same about Claude-SearchBot and Claude-User: disabling them "may reduce your site's visibility". Citation in AI answers depends on the search crawlers. If your robots.txt was written to block AI crawlers wholesale, check whether it also blocks the search crawlers that would cite you.
- How this plays into our Althea Labs audit: We analyze your server-level permissions to make sure you aren't inadvertently hiding yourself from the very systems that drive your sales pipeline, and that a training block has not turned into a search block by accident.
2. Using Schema as the Context Layer
AI models don't read your site like humans because they aren't humans. They're machines seeking code that tells them exactly what they are looking for, which is also what a prospective customer is querying them to find. If your product features, compliance standards, or pricing are buried in flowery marketing prose, the machine will try to approximate answers, and those guesses lead to hallucinations.
To make sure that an AI model can identify the necessary brand information, you need sign posts that point the way. Schema markup is one of them.
- The Reality: Schema is neither a requirement nor a guarantee. Google's AI features documentation says there are no additional requirements to appear in AI Overviews or AI Mode, and its guide to generative AI features lists overfocusing on structured data among the myths. We treat JSON-LD markup as hygiene, because it removes ambiguity about who you are and what you sell. There is also evidence that it travels with visibility. In "The Impact of JSON-LD Metadata on ChatGPT Visibility", an SSRN working paper by Peter Schanbacher, a study of 1,508 real estate agents in Germany found that 17.2% of the agents visible in ChatGPT used Product schema, against 1.8% of those who were not visible. That is a correlation, not proof that the schema caused the visibility, but it is a large gap.
- Question-and-answer pairs: FAQPage schema no longer earns rich results. In August 2023, Google limited FAQ rich results to well-known, authoritative government and health websites, and it stopped showing them altogether on 7 May 2026. We still recommend clear question-and-answer pairs on the page, in plain HTML with or without the markup, because in our experience a direct question followed by a direct answer is the easiest thing for an AI answer to lift.
- How this plays into our Althea Labs audit: We look for the gap. Is your site providing a machine readable fact sheet, or just a wall of unstructured text?
3. The Logic Layer: Semantic Hierarchy
AI models prioritize content that reveals clear, immediate relationships between ideas. They're expecting the story to read like a math problem with subject and problem presented, solution enacted, and result delivered. Unlike a human reader who might browse an entire article for a lyrical story and enjoy the journey regardless of how long it takes, an AI agent uses a retrieval process that, in our experience, gives most weight to what it encounters first.
- Put the answer first: Our working rule is that the sentence an engine is most likely to quote belongs in the first paragraph under the heading that asks the question, not at the end of a build-up. State the fact, then explain it. What Google actually says about page structure goes through the documentation.
- Semantic Clarity: Moving beyond keyword stuffing toward a clear hierarchy of H1 to H3 tags tells the machine exactly how your product solves a specific buyer's problem.
- How this plays into our Althea Labs audit: We evaluate how your pages are front-loaded. We find your most citable facts, such as your unique data, statistics, and direct answers, and position them where the machine actually looks for them.
The Compounding Effect of B2B Site Infrastructure
Technical optimization isn't just about today's chatbot response. This precision does more than clean up your backend. It builds a self-reinforcing loop in which early AI citations compound into category authority, so your brand becomes a default recommendation rather than an occasional one.
Coming Up in Part 3
In Part 3: The PR Footprint, we shift from your website to the rest of your footprint. We'll explore why LLMs cite Reddit and earned media ahead of your own marketing copy and how to ensure those external sources are pointing in your direction. Part 4: Community Presence then covers the conversation happening on Reddit, forums and review sites. If you would rather see how the four inputs fit together in a program, PR Plus is how we run them.
Ready to See Where You Stand?
Every Althea Labs engagement starts with an audit. We show you exactly where your brand stands in search, in AI, and in the communities where your buyers actually are, then show you what it would take to build from there.
CONTINUE READING
