The three disciplines
- SEO Search Engine Optimization
- The work that makes a site easy for search engines such as Google and Bing to find and understand, so that its pages appear among the first results when someone looks for what you offer. What counts is how the page is built, the words it uses, how fast it is, whether it reads well on a phone and which other sites cite it. It is the foundation of the other two: a site Google cannot read, an artificial intelligence cannot read either.
- AEO Answer Engine Optimization
- The work that makes a page fit to become the direct answer: the box at the top of Google, the AI Overview, the answer a voice assistant reads out. You get there by writing clear, short answers, organised as questions and answers, and declaring them to the machine with structured data.
- GEO Generative Engine Optimization
- The work that helps a brand get named by ChatGPT, Gemini, Perplexity or Claude when they write an answer. These systems read many sources and name very few: they reward text that can be extracted, declared sources, a brand recognisable beyond its own site and, quite simply, permission to read it.
How engines read a site
- Crawler
- The program an engine sends to visit sites, following links from page to page; Google's is called Googlebot. If it cannot read a page, as far as the engine is concerned that page does not exist.
- Indexing
- The moment when the engine, having read a page, keeps it in its archive, the index, and can show it in results. A page that has been read is not necessarily indexed: it can be left out as a duplicate, as nearly empty or because it is blocked.
- robots.txt
- A text file at the root of a site (for example yoursite.com/robots.txt) that tells crawlers which parts they may visit. A single wrong line can lock out Google or the crawlers of artificial intelligence. Source: Google Search Central
- Sitemap
- A file, usually sitemap.xml, listing the pages of a site that engines should know about, with the date each was last updated. It guarantees nothing, but it helps new pages get discovered quickly. Source: sitemaps.org
- Canonical URL
- The address a site declares as the official version of a page that can be reached at several addresses, for example with and without parameters. It stops the engine from splitting the page's value among its copies. Source: Google Search Central
- hreflang
- The markup that ties together the versions of the same page in different languages, so the engine shows each person the one in their language. The versions must declare each other: if one stays silent, the markup does not count. Source: Google Search Central
- Redirect
- The instruction that sends visitors automatically from one address to another. A 301 means moved permanently and carries the page's value with it; chains of redirects slow the visit down and scatter that value. Source: Google Search Central
- 404 error
- The page not found response. A link that leads to a 404 is a broken link: it disappoints whoever follows it and wastes the time crawlers spend on the site.
- HTTPS
- The encrypted version of the web's protocol, the one with the padlock in the browser bar. Today it is the bare minimum: browsers flag sites without it as not secure.
Labels for machines
- Structured data
- Labels in the code, invisible to readers, that tell the machine what it is reading: this is a company, this is a product with its price, this is an article with its date, this is a question with its answer. Without them, the engine has to guess. Source: Google Search Central
- Schema.org
- The shared vocabulary of structured data, created in 2011 by Google, Microsoft and Yahoo, soon joined by Yandex. It defines the kinds of things that can be described, such as Organization, Product, Article or FAQPage, and their properties. Source: schema.org
- JSON-LD
- The format Google recommends for writing structured data: a block of code kept apart from the page's text, which can be added and corrected without touching anything else. Source: json-ld.org
- FAQPage
- The Schema.org type that declares that a block of the page is made of questions and answers, so that engines and artificial intelligence can pick them up directly. Source: schema.org/FAQPage
- Entity
- A recognisable, distinct thing, a company, a person, a place, a product, that the engine can identify without ambiguity. An engine that knows Rossi is that particular company, with that site and those profiles, can cite it; one that only sees a common word cannot.
- sameAs
- The structured-data property that links a site to the official profiles of the same entity elsewhere: Wikipedia, LinkedIn, social networks. It helps machines understand that it is always the same company.
- Open Graph
- The labels that decide the title, description and image a page shows when someone shares it on LinkedIn, WhatsApp, Facebook or X. Without them, each platform picks for itself, often badly. Source: ogp.me
Speed and experience
- Core Web Vitals
- The three measures Google uses to judge the experience of visiting a page: LCP, INP and CLS. Google uses them among the signals by which it ranks results. Source: Google Search Central
- LCP Largest Contentful Paint
- How long it takes for the largest element of the first screen to appear, usually a heading or an image. Google considers it good under 2.5 seconds. Source: web.dev
- INP Interaction to Next Paint
- How quickly the page responds when you tap or click. It is good under 200 milliseconds. In 2024 it replaced the old FID. Source: web.dev
- CLS Cumulative Layout Shift
- How much elements move while the page settles: the button that jumps away just as you are about to tap it. It is good under 0.1. Source: web.dev
- Lighthouse and PageSpeed Insights
- Google's free tools that test a page in the lab and give it a score from 0 to 100 for performance, accessibility, best practices and SEO. PageSpeed Insights adds, when available, data collected from real visitors. Source: Chrome for Developers
Answers and artificial intelligence
- SERP Search Engine Results Page
- The results page of a search engine: the links, the boxes, the maps and, more and more often, an answer written by artificial intelligence at the top.
- Featured snippet
- The box at the top of Google's results that quotes an extract from a page as the direct answer to the question, with a link to the source. It cannot be bought: you get there by writing the answer in a form the machine can cut out. Source: Google Search Central
- AI Overview
- The AI-generated answer Google shows above the results, citing a few sources. Whoever is cited there takes most of the attention; the rest of the list scrolls by below, often without a click. Source: Google Search Central
- AI crawlers
- The programs that read sites on behalf of artificial intelligence, such as OpenAI's GPTBot, Anthropic's ClaudeBot and PerplexityBot. They can be let in or blocked with a line in robots.txt, and many sites block them without knowing it.
- llms.txt
- A text file at the root of a site that introduces it to artificial intelligence in a few orderly lines: what the site does and which pages matter. It is a convention proposed in 2024 and still rare, which is exactly why having one stands out. Here is ours. Source: llmstxt.org
- Citation
- When a generative engine names a brand or points to a page as the source of its answer. It is the new first place: very few names make it into a paragraph written by AI.
- E-E-A-T Experience, Expertise, Authoritativeness, Trustworthiness
- The yardstick Google's guidelines use to judge whoever writes and the site that publishes. It is built with real bylines, a credible about page, cited sources and consistency between what you say about yourself and what others say about you. Source: Google Search Central
- Backlink
- A link from another site to yours. Links from authoritative, relevant sites remain one of the strongest signals for search engines, and beyond your own site they also build the recognition that artificial intelligence rewards.
Bottega's method
- AI agent
- A program built on artificial intelligence that receives a goal, decides for itself where to look, pieces the clues together and reports what it found. At Bottega there are six, each with its own trade, and they work for the consultant, not in their place.
- The consultant
- The senior person who opens and closes every job at Bottega: they listen, choose what matters, sign the plan and verify the corrections. In the Florentine way, we call them the master.
- The first look
- The free analysis you start by pasting your site's address: within minutes, the three grades, SEO, AEO and GEO, and the problems found. No access to your site, no commitment.
- Source-code verification
- The check, after every correction, that the change really is in the page's code, citing the line. Not a tick on a spreadsheet: proof.
- Score and letter
- Each discipline receives a score from 0 to 100 and a letter from A to F, and under the score the areas that make it up. No average across the three: an average would hide the area doing worst.
- Impact and effort
- The two estimates attached to every recommendation: how much it can improve visibility and how much work it takes. Together they tell you where to start.
Want to see these words applied to your own site? The first look is free and takes a few minutes.
The first look, free