Glossary
AI search, in plain English.
Plain-English explanations of the words you'll see in an AI Spy report. If a term in your report is underlined, it's explained here.
What the numbers mean
- Mentions
- How many of the AI answers named your business. If your name appears in 11 of the 15 answers we collected, you have 11 mentions. Being named is the first step to being recommended.
- Citations
- When an AI answer lists a website as a source, that is a citation. Your citations are the answers that listed your own website. Being named is good; being cited means the assistant relied on your own site for its information.
- Average position
- Where your business usually appears in an answer's list, where 1 is first. Lower is better. Customers tend to pay most attention to the first few names.
- Sentiment
- How positively the answers describe a business, from 0 (negative) to 100 (very positive). 50 means neutral, such as being listed without any comment.
- Biggest threat
- The competitor most likely to take your place in AI answers. We rank competitors by how often they appear ahead of you (or when you are missing), how often they are named at all, and how often their own website is cited.
Your website: the basics
- Technical SEO
- The behind-the-scenes basics that let search engines and AI systems find and read your website: page titles, descriptions, being open to indexing, a sitemap, and access for crawlers. SEO stands for search engine optimisation.
- Meta description (page description)
- A one or two sentence summary of a page, stored in the page's code. Search engines and AI systems often use it to describe the page.
- Canonical tag
- A line of code that tells search engines which web address is the main one for a page, for when the same content can be reached at several addresses. It stops them treating copies as separate pages.
- Indexing and noindex
- Indexing means a search engine has stored a page so it can be found. A noindex instruction tells search engines to leave a page out, which is useful for private pages but harmful if it is set by mistake.
- robots.txt
- A small file on your website that tells automated visitors, such as search engines and AI crawlers, which parts of the site they may read.
- AI crawlers
- Automated programmes run by AI companies, such as GPTBot for ChatGPT and ClaudeBot for Claude, that read websites. If your robots.txt blocks them, those assistants cannot read your site.
- Sitemap
- A list of the pages on your website, in a file that helps search engines and AI systems find them all.
- llms.txt
- A new, optional file that gives AI assistants a short plain-English guide to your site and your most useful pages. It is quick to add and costs nothing.
Agent readiness
- Agent readiness
- A score for how easily AI crawlers can get into your website and how easily AI agents can use its forms. It covers crawler access, whether a firewall is turning AI visitors away, and whether your forms are labelled and described so an agent can complete them.
- AI agents
- AI tools that don't just answer questions but act for a person: filling in a contact form, booking, or requesting a quote. They can only use a website whose forms they can read and complete.
- WebMCP
- A new, optional way for a website to tell AI agents, in plain terms, what each of its forms does and what to put in each field. Without it, an agent has to guess. It is early days, so think of it as getting ready for what is coming.
- Firewall
- Security software in front of a website that turns away traffic it doesn't trust. Some firewalls turn away AI crawlers by default, even when the site's robots.txt welcomes them.
- "I'm not a robot" test
- A puzzle or tick-box, often called a CAPTCHA, that stops automated software submitting a form. It keeps spam out, but it also stops an AI agent acting for a real customer.
Structured data
- Structured data (schema.org markup)
- Code added to a page that spells out, in a standard format that machines understand, who you are, what you offer and what customers say about you. It is invisible to visitors, but search engines and AI systems read it. The standard is called schema.org.
- Business markup
- The part of structured data that describes the business itself: its name, address, logo, phone number and links to its social profiles. It gives search engines and AI systems a reliable fact sheet about you instead of leaving them to guess from your page text.
- FAQ markup
- Structured data that labels the questions and answers on a page, so machines can read them as questions and answers.
- Review markup
- Structured data that labels customer reviews and star ratings so machines can read them.
Content
- Natural language
- Writing that answers questions the way people ask them, in plain conversational words rather than slogans. AI assistants tend to quote text that answers the question directly.
- Media content
- A score for the images and video on your website: whether pages have them, and whether images carry alt text, which describes each picture to people and machines that cannot see it.
- Alt text
- A short description of an image, stored in the page's code. Screen readers read it out to people who cannot see the image, and search engines use it too.
The big picture
- GEO (generative engine optimisation)
- Improving how AI assistants find, understand and recommend your business. It is the AI-era cousin of SEO.
- AI assistants and web search
- Tools like ChatGPT, Claude and Gemini that answer questions in conversation. When we test them we switch their web search on, because that is how customers use them, and it makes them list the sources they used.