The Next AI Protocol Won’t Save Your SEO Strategy

Sep 03, 2026 02:00 AM - 1 hour ago 1

Recently, I recovered myself successful different speech that has go progressively familiar. A elder executive had received a informing from an AI visibility appraisal vendor that the institution was not sufficiently prepared for AI search. Among the recommendations was thing I person seen appearing much frequently: the institution needed an llms.txt file.

Suddenly, an emerging, and still-debated publishing format had go an executive concern. Someone now needed to find whether the proposal was valid, measure the imaginable impact, explicate why the institution had not already implemented it, and determine whether trading and engineering resources should beryllium redirected to reside it.

I precocious wrote astir this arena arsenic portion of what I telephone the AI FUD Tax. The costs of the individual proposal whitethorn beryllium comparatively small, but the organizational costs of many times responding to the latest outer AI audit tin go substantial. Every caller audit, vendor pitch, protocol, acronym, aliases competitory declare creates different information of questions astir whether the statement is falling behind.

The problem is not llms.txt, MCP, markdown, aliases immoderate peculiar technology. In fact, they do different things. Some thief machines observe information, immoderate supply replacement ways to correspond it, and others specify really systems speech aliases entree it. Lumping them together, technically, would beryllium inaccurate, but strategically, they create the aforesaid organizational temptation: treating the latest transportation system arsenic the solution alternatively than examining the underlying knowledge. The correction is treating each 1 arsenic though it represents a caller strategy.

After watching variations of this rhythm for decades, I judge we are erstwhile again focusing excessively overmuch attraction connected the format and not capable connected the accusation those formats are expected to communicate. Instead of instantly asking, “Do we request to instrumentality this?”, organizations should first inquire a much basal question:

“Do we person the knowledge required to support it?”

That favoritism becomes progressively important arsenic AI creates much ways for machines to devour organizational information. If the underlying knowledge is incomplete, fragmented, inconsistent, aliases trapped wrong individual departments, adding different machine-readable format does not lick the problem. It simply creates different spot to people the aforesaid limitations.

The organizations champion positioned to accommodate will not needfully beryllium those that instrumentality each caller protocol first. They will beryllium those that shape and govern their knowledge good capable that supporting the adjacent useful format becomes a publishing determination alternatively than different reconstruction project.

Decision Coverage Creates The Next Question

In the erstwhile article successful this series, I introduced Decision Coverage arsenic a measurement to measurement really wholly an statement has exposed the grounds AI needs to evaluate, compare, qualify, and confidently urge its products aliases services.

The thought emerged from a problem I judge galore organizations are opening to encounter. They whitethorn person an tremendous magnitude of merchandise accusation yet deficiency the grounds AI needs to support an existent customer decision. Specifications tin picture what a merchandise is, but they do not needfully explicate who it is due for, erstwhile it should beryllium recommended, really it compares pinch alternatives, aliases which trade-offs matter to different customers.

This becomes particularly evident erstwhile we deconstruct analyzable prompts. A petition for the “best” merchandise is seldom a azygous question. It is typically a operation of stated requirements, inferred preferences, constraints, comparisons, and eligibility criteria that collectively find which options make the cut.

Consider a customer asking for the champion family-friendly beachfront edifice successful Cancun. “Best” is not an property a edifice tin simply adhd to a page. The proposal whitethorn dangle connected beachfront access, family suitability, room configuration, amenities, price, availability, reviews, and different criteria inferred from the request. The AI must measure those conditions collectively earlier deciding which properties suffice for consideration.

Decision Coverage approaches the aforesaid problem from the organization’s side. Once we understand the variables influencing the decision, we tin find whether the statement has charismatic grounds to support them. If a captious criterion cannot beryllium substantiated, the problem whitethorn not beryllium that the marque classed poorly. It whitethorn ne'er person provided capable grounds to make the cut.

This gives america a much defensible measurement to diagnose a deficiency of AI visibility. Rather than watching that a competitor was recommended and instantly responding pinch much “me excessively parity content,” much links, aliases different analyzable method implementation, we should deconstruct the decision, place the criteria influencing qualification, and find wherever our supporting grounds is incomplete.

In different words, we tin statesman moving from simply watching what happened toward processing grounds for why it happened. Once we cognize what grounds should exist, however, Decision Coverage raises different question: Where should that knowledge reside, and really do we guarantee each strategy receives the aforesaid complete answer?

That is wherever the obsession pinch these caller individual formats originates to create problems.

A New Format Cannot Fix Missing Knowledge

The easiest consequence to a Decision Coverage spread is to put the missing accusation into immoderate format is receiving attraction astatine the moment. Daily LinkedIn is flooded pinch recommendations to grow the schema, create a markdown version, aliases build an MCP endpoint, and yes, deploy llms.txt.

That whitethorn lick an contiguous publishing problem, but it does not needfully lick the underlying knowledge problem. If a customer determination depends upon 5 meaningful criteria and the statement tin substantiate only four, publishing those aforesaid 4 pieces of grounds done different protocol does not abruptly found the fifth. We person simply made the aforesaid grounds spread disposable successful different format.

This favoritism sounds obvious, yet overmuch of today’s chat of AI implementation reverses the order, focusing connected citations alternatively than the sources aliases methods of integration. An expanded schema tin picture relationships betwixt entities, but it cannot find what those relationships should be. MCP tin make aggregate organizational resources accessible to AI systems, but it cannot find whether those resources incorporate the knowledge required to reply a customer’s question. An llms.txt record tin constituent machines toward information, but it cannot compensate for accusation the statement ne'er created.

All of these formats are mechanisms for communicating and transporting knowledge. Their worth yet depends connected the completeness and value of what organizations put into them. I person seen complete 100 agentic readiness audits each flagging the beingness of an llms.txt file, but nary of them critique the extent aliases value of the record for those that had them.

This is besides why the existent chat astir information integrity is truthful important. In his caller MCP article, Alex Moss based on that method SEO progressively needs to attraction connected maximizing information integrity arsenic AI systems dangle connected meticulous entities, definitive relationships, machine-readable formats, actions, and reliable cognition signals. He besides made an important constituent that aligns intimately pinch my statement here: Rather than trying to foretell which emerging protocol will yet win, organizations should fortify the underlying layers connected which those protocols depend.

Data integrity, however, besides creates a preceding organizational question. Before we tin guarantee that accusation remains accurate, synchronized, and trustworthy, we request to find what knowledge should exist, how those pieces subordinate to 1 another, who owns them, and which root should beryllium considered authoritative.

Data integrity helps guarantee that accusation remains trustworthy erstwhile it exists. Knowledge architecture helps guarantee that the correct knowledge exists, is connected appropriately, and tin beryllium governed arsenic an organizational asset. Both go progressively important arsenic AI systems trust connected our accusation to make decisions alternatively than simply return documents.

Build The Base Once, Publish Everywhere

During a caller webinar, I tried to simplify this progressively analyzable AI scenery into a azygous principle: Build the canonical guidelines once. Publish everywhere.

Diagram showing knowledge built erstwhile astatine a canonical guidelines and published everywhere

The thought is intentionally elemental because organizations do not request different acronym aliases protocol to manage. They request to extremity rebuilding the aforesaid knowledge for each caller destination.

At the instauration are the facts, relationships, policies, expertise, customer determination criteria, and supporting grounds the statement tin authoritatively provide. Those elements request to beryllium captured successful a canonical knowledge root wherever they tin beryllium connected, governed, updated, and reused independently of immoderate peculiar publishing format. AI-focused organizations for illustration Milestone person built this natively into their contented guidance strategy arsenic the canonical root of truth, tin of outputting to immoderate existent and emerging markup formats.

Decision Coverage helps america find whether that knowledge is sufficiently complete to support the customer decisions that matter. Knowledge architecture provides a measurement to shape and support knowledge. The various formats past go transportation mechanisms for making the due portions of that knowledge disposable to the systems that request them.

Today, those transportation mechanisms whitethorn see web content, schema markup, Merchant Center feeds, APIs, markdown, MCP, llms.txt, and different emerging formats. Tomorrow, the database will almost surely beryllium different.

That should not require rebuilding the underlying knowledge.

If each publishing format maintains its ain type of the organization’s facts and determination evidence, each update creates synchronization problems and each caller protocol creates different implementation project. If those formats tie from a common, governed knowledge source, the architecture changes dramatically. The difficult activity happens erstwhile astatine the knowledge layer, while the publication furniture adapts arsenic technologies and requirements change.

This is what I mention to arsenic Knowledge Architecture. It is the organizational capacity that allows trusted knowledge to beryllium consistently assembled, governed, and delivered wherever it creates value.

Publication Is Not Capability

This favoritism besides explains immoderate of my vexation pinch the progressively loose usage of position specified arsenic “AI ready” and “agentic ready.”

Supporting an agent-oriented protocol tin surely make it easier for an supplier to interact pinch an organization’s systems. That is useful infrastructure, but it does not automatically mean the statement possesses the knowledge required for an supplier to make a useful decision.

There are really 2 different problems being conflated. The first is organizational capability: Can the institution capture, connect, govern, maintain, and retrieve the knowledge customers and machines need? The 2nd is publication: Can that knowledge beryllium expressed done the format required by a peculiar hunt engine, AI platform, agent, application, aliases protocol?

One of my arguments astir these markup formats is that they travel from a vendor’s solution that promises a intelligibly defined deliverable. They are solving a existent problem for AI systems: the chaos of a modern marketing-focused website, and that makes it easy to sell. Some of their devices expound connected really quickly and easy they tin rotation retired the format, pinch small to nary mention of really they tin wrangle organizational knowledge, since nary protocol tin create it for us.

A protocol cannot reconcile conflicting merchandise accusation owned by different departments. It cannot extract the expertise surviving wrong a salesperson’s head, find which customer objections matter, found really a argumentation affects a peculiar product, aliases create the missing grounds identified done Decision Coverage. Those are organizational knowledge problems that must beryllium resolved earlier exertion tin administer the answers.

Technology tin expose organizational capability, but it cannot substitute for it.

This is why organizations do not go AI-ready simply by implementing much protocols. They go AI-ready by organizing their knowledge good capable that each useful protocol becomes different publishing destination alternatively than different effort to reconstruct what the statement knows.

Organize Knowledge Around Decisions

For astir of the web’s history, the attraction was connected creating and managing collections of pages; CMS platforms reinforced that structure, and SEO people optimized merchandise pages, class pages, articles, FAQs, and landing pages because pages were the superior units done which hunt engines retrieved accusation and customers consumed it.

AI is weakening that relationship.

A azygous customer mobility whitethorn now origin an AI strategy to retrieve accusation from aggregate pages, merchandise feeds, system data, reviews, outer sources, databases, and different knowledge repositories earlier synthesizing a response. The targeted webpage remains important, but it is nary longer needfully the portion astir which the determination is constructed.

The much durable organizing rule is the customer decision.

What does the customer request to know? Which criteria find whether a merchandise qualifies? What grounds supports those criteria? Which alternatives request to beryllium compared? What trade-offs should beryllium understood? Which policies, locations, readiness constraints, aliases different relationships impact the outcome?

These questions link straight backmost to the Decision Coverage framework. If we understand the conditions that collectively find whether a product, service, aliases statement makes the cut, we tin place the grounds required to support each condition. The canonical knowledge root past provides a governed location for that evidence, while individual formats find really it is delivered.

This is why chasing output formats gets the series backward. We should not statesman pinch an quiet protocol and inquire what accusation we tin put into it. We should statesman pinch the customer decisions we request to support, guarantee the grounds required for those decisions exists, and past find which transportation mechanisms make that knowledge accessible to the systems influencing the outcome.

This Is Infrastructure, Not Another AI Tactic

This statement should sound acquainted to anyone who has followed my erstwhile MCP articles because it extends a position I person held for immoderate time. In “SEO Is Not a Tactic. It’s Infrastructure for Growth,” I based on that sustainable hunt capacity depends connected capabilities embedded crossed the statement alternatively than isolated optimization projects. Search useful champion erstwhile product, content, technology, and business strategy are connected astir really customers really discover, evaluate, and take solutions.

I later introduced the Search Equity Gap to quantify the business worth mislaid erstwhile organizations neglect to seizure the qualified visibility they should reasonably earn. That article besides highlighted the increasing effect of zero-click experiences and AI-mediated search, successful which visibility tin stay moreover arsenic the economics of customer interactions change.

Those ideas person go progressively connected arsenic AI has evolved. Brand Sovereignty establishes the request for organizations to go the charismatic root of their ain facts and expertise. Click Worthiness helps find wherever continued engagement yields capable incremental customer and business worth to warrant investment. Decision Coverage asks whether we person exposed the grounds AI needs to evaluate, compare, qualify, and confidently urge us.

Knowledge architecture provides the durable instauration underneath those capabilities.

Without trustworthy organizational information, marque sovereignty becomes difficult to establish. Without knowing customer decisions, it becomes difficult to prioritize click worthiness. Without complete determination evidence, Decision Coverage remains incomplete. Without an architecture tin of governing and reusing that knowledge, each caller AI format sends the statement backmost into different implementation cycle.

This is why the adjacent AI protocol will not prevention an SEO strategy that lacks the underlying knowledge capability. The protocol is downstream of the problem.

The Next Protocol Won’t Be The Last

I do not cognize whether MCP, llms.txt, Agents.md, UCP, aliases immoderate of today’s different emerging approaches will go foundational standards. Some undoubtedly will go much important. Others will evolve, merge into thing else, aliases vanish arsenic quickly arsenic erstwhile technologies that erstwhile seemed essential. That uncertainty is precisely why organizations should defy building their AI strategy astir individual formats. The adjacent protocol will not beryllium the past one.

Organizations that build their strategy astir today’s hyped formats will yet request to rebuild astir tomorrow’s. Organizations that build a governed knowledge root astir their business expertise and customer decisions will beryllium successful a very different position. When the adjacent useful format appears, they will not request to rediscover, recreate, and reconcile the knowledge it requires. They will request to find whether the format provides a valuable caller measurement to people what they already know.

That is simply a overmuch much resilient exemplary for AI readiness and, much importantly, a overmuch amended usage of organizational resources.

The extremity is not to support each imaginable format. It is to guarantee that the knowledge required to make important customer decisions is complete, authoritative, governed, and disposable for transportation done the formats that matter. The protocols will proceed to change. The transportation mechanisms will proceed to multiply, and consultants will undoubtedly proceed uncovering caller items to adhd to AI readiness audits. The strategical consequence should stay remarkably stable.

Build the guidelines once. Publish everywhere.

More Resources: