What Is AI Distribution?

AI Distribution is how AI systems decide which businesses get customers. We scored 1,431 businesses across 16 US cities. 72% are invisible.

A patient asks ChatGPT for a dentist in Phoenix. A bride asks Gemini for a med spa in Austin. A founder asks Perplexity for a law firm in Chicago. Each AI recommends three businesses. Hundreds are ignored.

AI Distribution is how AI systems decide which businesses get customers. Not which businesses appear in a list. Which businesses get recommended, chosen, and acted on when a person asks an AI for help.

This is not a new marketing channel. It is a new allocation mechanism. AI systems evaluate every business in a market, select the ones they trust, and send real customers to them. The businesses not selected don't receive a notification. They simply stop being part of the conversation.

Visibility Is the Wrong Word

The industry settled on "AI visibility" to describe this shift. The term undersells the problem by an order of magnitude.

Visibility is passive. A billboard is visible. A directory listing is visible. Visibility means existing in a space where someone might look. It says nothing about whether anyone actually walks through your door.

AI systems don't make businesses visible. They distribute customers. When someone asks an AI assistant for a recommendation, the model evaluates, selects, and presents. By the time the person sees the answer, the choosing is already done. They arrive pre-decided. Microsoft's 2025 study of 1,200+ sites found that customers arriving through AI recommendations convert at three times the rate of traditional search traffic. The AI already did the choosing. The customer is following through.

The distinction matters because it changes what you measure. Visibility measurement asks: does your business appear? Distribution measurement asks: is the system actively sending customers to you, and if so, which types of customers, on which platforms, for which needs?

You are not competing for attention. You are competing for allocation.

The Five Dimensions

AI Distribution follows a dependency chain. Five dimensions, each requiring the one before it.

D1: Readable. Can AI parse your business at all? Structured data, schema markup, machine-readable content, crawl access. If your technical foundation blocks AI systems from reading your information, nothing downstream matters. 93.4% of the businesses we scored are missing basic machine-readability signals.

D2: Findable. When someone asks a relevant question, does AI surface your business? This is where every existing AI visibility tool stops. They check if you show up. But showing up once for a generic query is not distribution.

D3: Credible. Does the evidence support the recommendation? AI systems don't evaluate your website the way a human would. They evaluate what the rest of the internet says about you. Reviews, citations from independent sources, cross-platform consistency, published work. Third-party validation, not website quality, drives AI recommendations. AI doesn't visit your website. It reads your reputation.

D4: Transactable. Can a customer act on the recommendation without friction? If AI recommends your business but the customer can't book, request a consultation, or complete a purchase, the distribution chain breaks. In January 2026, Google launched the Universal Commerce Protocol. In February, OpenAI launched Instant Checkout in ChatGPT. The infrastructure for AI agents to complete transactions is live. Businesses without machine-readable pricing, booking capability, and payment infrastructure are invisible to this entire layer.

D5: Irreplaceable. Is your business distinct enough that AI cannot substitute another? When every competitor offers the same service described the same way, AI has no reason to prefer one over another. When a dental practice publishes its implant success rates, or a law firm develops a named case evaluation methodology, AI has a reason it cannot find elsewhere. The test: can AI generate a functionally equivalent alternative? If yes, it is not D5.

The chain is structural, not arbitrary. A business with broken schema markup cannot be parsed by AI crawlers, regardless of its reputation. A business with no AI presence cannot accumulate the cross-platform citations that build credibility. Each dimension creates the precondition for the next.

Where Existing Approaches Stop

GEO. AEO. AI SEO. LLM Optimization. The AI marketing landscape has filled with acronyms. Each describes a real tactic within D1 and D2: structuring content for machine extraction and earning visibility in AI responses.

Valuable work. Partial coverage. Every existing approach addresses a real problem within the first two dimensions of a five-dimension system. Three of five dimensions are unmeasured by every other tool in the market. These are the dimensions that determine whether a recommendation converts and whether your position holds.

The relationship is containment, not competition. AI Distribution is not a tactic. It is the full scope of how AI systems evaluate, recommend, and route customers to businesses. GEO improves D1 and D2. A credibility strategy improves D3. Transaction infrastructure improves D4. Differentiation strategy improves D5. AI Distribution is the measurement of all five.

What 1,431 Businesses Revealed

We scored 1,431 businesses across 16 US cities using The AI Craftsman's Reverse Engineering Protocol, which derives scoring criteria empirically from AI recommendation engines rather than theorizing what they value. All data as of Q1 2026.

72% are functionally invisible. When AI is asked about their market, their name does not appear. Independent research from SOCi across 350,000 business locations confirms the pattern at scale: ChatGPT recommends just 1.2% of local businesses. Perplexity, 7.4%. Even Gemini, the most generous, selects only 11%. The other 98.8% don't get a rejection letter. They simply don't exist in the conversation.

Website quality does not predict AI recommendations. The correlation coefficient between website quality and AI citation frequency: 0.02. Near zero. Across four US markets, 114 practices, replicated four times. The practices with the most AI citations often had the weakest websites.

Consider what this means for a practice that just invested $15,000 in a website redesign. The investment improves patient experience. It does not measurably improve AI distribution. The signal AI rewards is external validation: what reviews, professional directories, and independent sources say about your business carries more weight than anything on your website.

Different platforms recommend different businesses. Only 11% of businesses cited by one AI platform are also cited by another (Profound, 680M citations analyzed). If your business leads traditional Google search, there is a 55% chance it does not appear in AI recommendations at all (SOCi, 350,000+ locations, 2026). A business dominating on ChatGPT may be invisible on Gemini. Single-platform measurement captures a fraction of reality.

The gap compounds. Research on AI recommendation systems confirms they create their own champions: businesses already recommended get recommended more often (Calvano et al., Oxford Review of Economic Policy, 2024). The data trains the model, the model produces recommendations, the recommendations generate more data. Every month without measurement is a month the distribution gap widens quietly.

Where Do You Stand?

The simplest test takes 30 seconds: ask ChatGPT, Gemini, and Perplexity for your service in your city. Note which ones mention you. That is your cross-platform distribution snapshot.

If you appear on one platform but not the others, you have visibility, not distribution. If you appear for broad queries but vanish when the question gets specific ("dentist for nervous patients in Phoenix" instead of "best dentist in Phoenix"), your distribution is thinner than it looks.

If you don't appear at all, the gap is structural. No amount of ad spend or traditional SEO will solve a machine-readability problem. The fix starts at D1 and builds upward.

The question most businesses have never asked is the one that matters now: not "how do I rank?" but "when AI decides who gets the customer, is it choosing me?"

The AI Distribution Score measures exactly that. A 0-100 composite across all five dimensions, with per-dimension sub-scores that reveal where the chain holds and where it breaks. Check yours at theaicraftsman.com.

The Compounding Window

Credit scores existed before FICO standardized them. Website authority existed before Moz created Domain Authority. The phenomenon was real. The measurement standard was missing.

AI Distribution is at that inflection now. 45% of consumers now use AI to find local services. One year ago: 6% (BrightLocal, 2026). That is not a trend line. That is a phase transition. The distribution is happening whether you measure it or not.

The businesses that establish their AI Distribution profile now build compound advantage that late movers cannot replicate. Not because early measurement is inherently valuable. Because AI recommendation systems reinforce what they already know. Every month of measurement creates a baseline that cannot be generated after the fact. Start measuring in March, and someone starting in September can never recover those six months of competitive intelligence.

The cost of waiting is not stagnation. It is falling behind a curve that accelerates without you.