The question we're answering
When someone asks ChatGPT, Perplexity, Gemini, or any AI assistant to recommend a business like yours, can it? Does it have enough structured, verifiable information to confidently include you in the answer?
The AO score measures how close your business is to being consistently recommendable by AI agents. A higher score means more data, better structured, with fewer gaps for an agent to stumble on.
Six dimensions, weighted by impact
Each dimension measures a distinct aspect of AI readability. The weights reflect how much each dimension affects whether an AI agent can actually recommend you. These weights are based on how current AI systems (as of early 2026) retrieve, evaluate, and present local service businesses.
What it measures: Whether AI agents can read your business information directly, or whether they're guessing from your page text.
This is the single most impactful dimension. When your business data is properly structured, AI agents get clean, unambiguous facts: what you do, where you are, what you charge, what credentials you hold. Without it, they're scraping text and hoping for the best. This is the foundation of algorithmic trust.
- Can AI agents parse your business information directly?
- Are your business type, services, and location clearly defined for machines?
- Are key details present: name, description, address, phone, services, pricing?
- Is your data properly connected so agents understand how your services relate to your business?
What it measures: Whether AI agents can extract a clear, specific list of what you actually do.
A vague "we offer a range of services" is invisible to an agent filtering by specific service type. This dimension checks whether your services are individually named, described, and structured so an agent can match them to what someone is looking for.
- Each service individually listed and described for AI agents
- Service names that match how people actually search ("boiler repair" not "heating solutions")
- Service descriptions with enough detail to differentiate from competitors
- Coverage of expected services for your industry profile
What it measures: Whether an AI agent can answer the question "how much does this cost?" without guessing.
AI agents increasingly filter by budget. If someone asks for "a plumber under 100 for a callout", businesses without any pricing signal are excluded entirely. You don't need to list every price. A starting rate, a price range, or structured Offer data is enough to stay in the conversation.
- Pricing data that AI agents can read and compare (not just text on a page)
- Visible pricing on the website (starting rates, price ranges, rate cards)
- Whether pricing is "contact for quote" only (reduces score significantly)
- Consistency between AI-readable prices and what's visible on your page
What it measures: Whether there's enough verifiable evidence for an AI agent to recommend you with confidence.
AI agents cross-reference multiple sources. They look for consistency (does the name, address, and phone match across your website and Google?), evidence of quality (reviews, accreditations), and whether your site actively allows AI crawlers to index it.
- Consistency: does your name, address, and phone match everywhere AI looks?
- AI crawler access: can ChatGPT, Claude, and Perplexity actually visit your site?
- Online presence: are your social profiles and directories connected to your website?
- Accreditations and qualifications mentioned on-site
- Review signals and what customers are saying about you
What it measures: Whether an AI agent can determine where you operate, when you're available, and how to book.
An agent that knows you exist but can't tell if you serve a given area, what your hours are, or how to book is unlikely to recommend you over a competitor with clearer data. This dimension checks the practical "can I actually use this business?" signals.
- Can AI agents determine exactly where you operate?
- Is your service area clearly defined?
- Are your opening hours readable by AI, not just visible on the page?
- Can an AI agent figure out how to book or contact you?
What it measures: What AI systems currently know about your business when asked directly.
We query major AI models with natural-language questions about your business and compare their responses to your actual details. This is the reality check: regardless of what data exists on your site, does the AI actually know who you are?
- Does ChatGPT/GPT-4 know your business name, location, and services?
- Accuracy of AI-generated descriptions vs. your actual offerings
- Whether the AI confuses you with competitors or generates incorrect details
- This module is optional and noted in the report if skipped
How the score works
Each dimension is scored 0-100 individually, then multiplied by its weight. The six weighted scores are summed to give your overall AO score out of 100.
Your score places you in one of five bands:
| Band | Score | What it means |
|---|---|---|
| Invisible | 0 - 25 | AI agents cannot confidently evaluate or recommend your business |
| Fragmented | 26 - 50 | Some signals present but major gaps remain |
| Visible | 51 - 75 | AI agents have some data but can't confidently recommend your services |
| Ready | 76 - 90 | Well-positioned for confident AI agent recommendation |
| Optimised | 91 - 100 | Fully agent-ready across all dimensions |
Why these dimensions?
These aren't arbitrary categories. They map to the actual process an AI agent follows when deciding whether to recommend a business:
What we don't score
The AO Audit is specifically about AI agent readability. It does not measure:
- Website design or visual quality
- SEO rankings or keyword performance
- Social media following or engagement
- Business quality or customer satisfaction
A beautiful website with no AI-readable data will score poorly. An ugly website with excellent algorithmic trust signals will score well. That's by design: we're measuring what machines can read, not what humans see.
Verifiable outputs
When we implement AO recommendations, every change produces verifiable output. The data we add to your website can be tested using Google's Rich Results Test and the Schema Markup Validator. You can paste your URL into either tool and see exactly what AI-readable information exists, whether it's valid, and what it contains.
This isn't a trust-me deliverable. It's testable, standards-based, and visible to anyone who checks.
Industry profiles
Not all businesses need the same data. A therapist's ideal profile looks different from a plumber's. The audit uses industry-specific profiles that adjust expectations: which data points matter most, which accreditations are relevant, which services a business of your type would typically offer.
Current profiles: therapist/wellness, interpreter/language services, and a generic profile that works for most local service businesses. We're building more as we audit more industries.
What's coming next
The current audit focuses on your website and on-site signals. But AI agents don't just read your site — they cross-reference multiple sources. We're expanding the methodology to include off-site signals that increasingly affect whether you get recommended:
- Third-party mentions: Are authoritative sites, directories, or publications citing your business?
- Community presence: Do you show up in Reddit, Quora, YouTube, or forums where AI models are trained?
- Review sentiment: Not just star ratings — does the language in your reviews match what people ask AI? ("gentle sound bath for anxiety" matters more than "5 stars great")
- Brand narrative protection: Is incorrect information about your business circulating online? Defensive AO.
- Customer language mirroring: Does your site use the phrases real customers use, or industry jargon they'd never type?
These signals will be integrated into the audit as we validate their impact through real-world testing. We'd rather add dimensions that are proven to matter than pad the score with guesses.
Limitations and honesty
This methodology is version 1. Here's what you should know:
- The weights are our best assessment of current AI agent behaviour, not a universal standard
- AI agent technology is evolving rapidly - what matters most may shift as systems change
- The AI Visibility module queries GPT-4 specifically; other models may have different knowledge
- We cannot guarantee that improving your score will result in AI recommendations - we can only ensure you have the data that makes recommendations possible
- We update the methodology as we learn more from real audits and as AI systems evolve
We'd rather be transparent about the edges than pretend they don't exist.
See where your business stands
We'll audit your website and send you your score within 2 working days. Free, no obligation.
Get your free AO score →Already know you need it?
Get the full AO Audit (£247) →