What is GEO?

Generative Engine Optimization is the practice of making your website readable, trustworthy and quotable for AI answer engines such as ChatGPT, Perplexity, Google AI Overviews, Gemini, Claude and Copilot, so that when a buyer asks those engines a question in your category, your company is the source they cite or recommend. Where SEO optimizes for a ranked list of links, GEO optimizes for being the answer itself.

SEO earns a position on a results page; GEO earns a citation inside a generated answer. The inputs overlap (good content, clean technical health) but the priorities shift. AI engines care whether their crawlers are allowed in, whether your pages carry structured data they can parse, whether your content is organized as direct answers with clear headings, and whether there are trust signals such as named authors, dates and outbound citations. A page can rank well in Google and still never be cited by ChatGPT.

Today the engines that drive measurable traffic and influence are ChatGPT (including ChatGPT search), Google AI Overviews and AI Mode, Perplexity, Gemini, Claude and Microsoft Copilot. Each uses its own crawler: GPTBot and OAI-SearchBot for OpenAI, Google-Extended for Google's models, PerplexityBot, ClaudeBot, and Bingbot for Copilot. A GEO audit checks that each of these can reach your site.

llms.txt is a plain-text file published at the root of your domain (yoursite.com/llms.txt) that gives AI systems a concise, structured summary of who you are, what you offer and which pages matter most. It is to large language models what robots.txt and sitemap.xml are to search crawlers. Most companies do not publish one yet, which makes it a quick win. Protony publishes its own at protony.ai/llms.txt.

It keeps your content out of training data, but it also removes you from AI answers. Several large brands in our AI Visibility Index block GPTBot or ClaudeBot and, as a result, cannot be cited when a buyer asks those engines for a recommendation. The usual answer for a commercial site is to allow the search-oriented crawlers (OAI-SearchBot, PerplexityBot, Google-Extended) while deciding deliberately about training-only bots.

Schema.org JSON-LD tells a machine exactly what a page is about: an Organization, a Product, a Service, an FAQ, an article with an author and a date. AI engines use these signals to decide what a page represents and whether to trust it. In our index, missing JSON-LD was the single largest point loss across brands, costing up to 14 of 20 points in the structured data category.

Technical fixes (crawler access, llms.txt, schema, metadata) are picked up within days to a few weeks as engines recrawl. Content restructuring and authority signals take longer, typically one to three months, because they depend on recrawl cycles and on the engine's confidence in your site. We measure monthly by running the same prompts across engines and recording who gets cited.

Three ways. First, your GEO score, which measures whether your site is technically and structurally readable by AI. Second, citation share: for a fixed set of buyer prompts in your category, how often each engine names you versus competitors. Third, referral traffic and leads from AI engines, which now appear as distinct sources in analytics. We report all three.

Any company whose buyers research before they buy: B2B software and services, professional services, healthcare, home services, specialty retail. If a prospect might ask ChatGPT or Perplexity who to use, you need to be the answer. Protony focuses on mid-market B2B brands and growing service businesses that want this handled without hiring an in-house team.

Protony offers three fixed-scope packages: a GEO Audit, a Fix Sprint that implements the audit, and an ongoing Visibility Program. Pricing depends on the size of your site and how many pages need restructuring. Request a quote on the contact page and you will have a number within one business day.