We scanned 500,000 service businesses across 183 verticals and 395 cities. The results reveal how unprepared most businesses are for the agent economy.
Enter your email to unlock the complete March 2026 report with data from 500,000 businesses, dimension breakdowns, vertical comparisons, and city-level analysis.
No spam. Unsubscribe anytime.
The first comprehensive benchmark of AI Agent Preference across service businesses paints a clear picture: the overwhelming majority of service businesses are not ready for AI agents.
Of the 500,000 businesses we scanned, 29% of those with complete scores fell into the Agent Incompatible band (below 10 out of 100) on the AI Agent Preference Score. The average score across all businesses was 33 out of 100. Not a single service business in our dataset reached Agent Preferred status.
The gap is not about awareness or willingness. It is structural. Service businesses built websites for human visitors. They have phone numbers, contact forms, and photo galleries. None of that infrastructure is readable by AI agents. The businesses are invisible to the systems that are increasingly making purchasing decisions on behalf of consumers.
Note: The national average includes businesses across all 395 cities in our database, including smaller markets where certain industry verticals score higher. Major metros tend to be competitive, densely scored markets where individual scores cluster below the national mean.
The headline finding: 29% of fully scored service businesses are Agent Incompatible. AI agents cannot book, quote, or pay the vast majority of service providers. These businesses functionally do not exist in the agent economy.
Scores map to capability bands: Agent Preferred (90-100), Agent Optimized (70-89), Agent Ready (50-69), Agent Functional (30-49), Agent Detected (10-29), Agent Incompatible (0-9). Here is how fully scored businesses distributed:
Percentages reflect fully scored businesses. Businesses with incomplete scans are excluded.
The distribution is heavily skewed toward the bottom. 29% of fully scored businesses fall into the Agent Incompatible band. The Agent Detected businesses (35%) typically have some structured data deployed but lack transactional capabilities. The handful reaching Agent Functional or Agent Ready tend to be practices using modern scheduling platforms that expose some structured availability data.
Performance varies by industry, but every vertical we measured averages Agent Incompatible.
| Vertical | Businesses | Avg Score | Band | % Agent Incompatible |
|---|---|---|---|---|
| Dental | 3,400 | 44 | Agent Functional | 19% |
| Legal | 3,100 | 42 | Agent Functional | 21% |
| HVAC | 3,700 | 42 | Agent Functional | 19% |
| Cleaning Service | 3,600 | 35 | Agent Functional | 31% |
| Plumbing | 1,700 | 39 | Agent Functional | 23% |
| Auto Repair | 3,700 | 37 | Agent Functional | 25% |
| Roofing | 2,600 | 39 | Agent Functional | 24% |
| Electrical | 4,800 | 35 | Agent Functional | 31% |
| Pest Control | 3,500 | 37 | Agent Functional | 26% |
Showing 9 of 183 verticals in our dataset, ordered by average score.
Dental and legal practices lead the pack, largely because patient portal and intake platforms provide some structured booking data. But even they average only 44/100, placing them in the Agent Functional band. Trade services (plumbing, roofing, electrical) cluster at the bottom. Their websites are built almost entirely for human visitors with no machine-readable infrastructure.
The four dimensions reveal where the biggest gaps are across all businesses:
Competitive Position is calculated individually for each business based on their local peer group. It is not included in aggregate benchmarks.
Agent Accessibility (41/100) is the highest-scoring dimension because most businesses have basic website structures that agents can partially navigate, even if they are not optimized for automated interaction. Data Reliability (30/100) benefits from businesses that maintain some consistent information across the web, though accuracy issues are widespread.
The transactional dimension tells the real story. Transaction Completeness averages 27/100. Almost no service businesses have agent-accessible booking, quoting, or payment flows. Structured pricing, live availability, and machine-readable checkout are functionally nonexistent across the benchmark.
The pattern is clear: Businesses have partial infrastructure for agent navigation (Agent Accessibility, Data Reliability) but near-zero infrastructure for agent transactions and protocol connectivity (Transaction Completeness). The gap between "navigable" and "operable" is where the entire AAO opportunity lives.
Five anonymized businesses from the dataset that illustrate common patterns:
This practice uses a patient portal with Zocdoc integration, giving it above-average Transaction Completeness. But limited agent protocol adoption and data structure keep the overall score in the Agent Incompatible band.
Typical trade service pattern. Has a basic website that agents can partially navigate but zero transactional infrastructure. An AI agent can crawl the site but cannot book, quote, or pay.
Better than average due to a professional website with structured contact pages and practice area descriptions. But no booking API, no payment processing, and service descriptions are in prose rather than structured data.
Has a quote request form, giving it a non-zero Transaction Completeness score. But zero agent protocol adoption and limited operational data structure keep the score low.
A rare Agent Functional business. Uses a modern platform with online booking, structured pricing, and partial payment processing. Has schema.org markup deployed correctly. This is what the path toward Agent Ready looks like and where most businesses should aim to get.
The data tells a simple story. Service businesses built their digital presence for humans. They have not rebuilt it for AI agents. The infrastructure gap is not 10% or 20%. It is nearly total.
Businesses in the Agent Incompatible band (0-9, the majority) have effectively zero agent infrastructure. They are invisible to the systems that are increasingly mediating how consumers find, compare, and book services.
Businesses in the Agent Detected (10-29) and Agent Functional (30-49) bands have some foundation to build on, typically basic structured data and a contact form. These are the businesses closest to becoming Agent Ready with targeted improvements.
Businesses reaching Agent Ready or above (55+) have meaningful agent infrastructure. They tend to use modern platforms with built-in structured data and online booking. They are the early winners in the agent economy.
The opportunity: Because the bar is so low, even basic AAO improvements create significant competitive advantage. Adding foundational AI agent infrastructure can significantly improve a business's AI Agent Preference Score. In a market where 29% are Agent Incompatible, reaching Agent Functional is a competitive moat.
Our scoring engine analyzes multiple facets of a business's public web presence across four dimensions of AI agent readiness.
Each of four dimensions is scored 0-100 and combined to produce the AI Agent Preference Score. Businesses were discovered through public search engines, business directories, and direct URL submission. The dataset includes businesses across 130+ service verticals and 390+ US cities.
All scores reflect the state of each business's web infrastructure at the time of scanning (March 1-17, 2026). Scores change as businesses update their infrastructure. Individual business scores are available through the free scan tool.
Get your free AI Agent Preference Score across all 4 dimensions. See how you compare to these benchmarks. 60 seconds. No credit card.
Get Your Free ScoreWant the full report as a PDF? Get your free scan and we will send you the complete March 2026 benchmark data. | What is AI Agent Preference? | Glossary