31 terms, from GEO to open territory. The ones Periscopy measures show where to read them in the product.
AI ads
Paid placement inside the answers of AI assistants, next to organic presence. On Google, ads enter AI Overviews and AI Mode automatically from Search, Shopping and Performance Max campaigns. ChatGPT shows ads on its free plans (the United States and the United Kingdom first, the European Economic Area from August 2026). Claude and Perplexity carry no ads, by their own decision. Measuring paid and organic separately is what stops an ad being read as an earned citation.
Answer Engine Optimization
A practical synonym for GEO. The focus is the same: appearing in answers generated by engines that synthesise, rather than in lists of links. Some markets say AEO, others say GEO. This site uses GEO as the umbrella term.
A purchase started or completed by an AI assistant on the user's behalf. The agent reads price, stock and delivery time in the brand's structured data (product feeds, the schema.org Offer) and can close the transaction inside the conversation. In August 2026, in-chat checkout was live in the United States (Google, Microsoft Copilot, Perplexity) and not yet in Europe, while the discovery layer (product panels in Google AI Mode, ChatGPT carousels) already worked in Portugal. A brand that is not in the data does not exist for the agent.
In Periscopy: CommerceThe robots assistants use to visit a site, such as GPTBot, ClaudeBot or PerplexityBot. Set against visibility, they show the machine reading the site before it cites it.
In Periscopy: TrafficThe conversational mode of Google Search, generalised in 2026. It reached one billion monthly users in May 2026. It progressively replaces the list of ten blue links with generated answers carrying three to five citations. In May 2026 its default model was Gemini 3.5 Flash. It is one of the surfaces Periscopy measures, always live, because that is how a person sees it.
In Periscopy: Engines and modesThe AI-written answer block at the top of a Google results page. It is measured live, like the other surfaces.
In Periscopy: Engines and modesAn AI usage policy file at the root of a site, similar in spirit to robots.txt. It states which AI crawlers may index the site and on what terms, covering GPTBot, ClaudeBot, Google-Extended, PerplexityBot and others.
The brand's average place in the answers that name it. It is read next to the citation rate, share of voice and share of recommendation.
In Periscopy: Engines and modesAn answer chooses a brand when it gives it a place in an ordering, a verdict against a criterion that names a single brand, or the closing sentence. Belonging to a set (“the big three”) does not count. When in doubt, it does not count.
In Periscopy: Chosen or mentionedThe rate at which a brand is mentioned in an engine's answers across a fixed set of questions relevant to its category. It is the primary GEO metric, expressed as a percentage (answers that mention the brand, divided by all answers, times 100). In each answer, it is a yes or a no.
In Periscopy: Chosen or mentionedHow machines identify the brand: Wikidata, Knowledge Panel, schema and sameAs. It is the layer that decides whether models fix the brand in memory.
In Periscopy: EntityThe work of establishing a brand as an entity in the knowledge graph: a consistent description across every platform (LinkedIn, Wikipedia, Crunchbase, G2), sameAs in the schema, and a clear role in the sector. It raises the probability of being cited by language models.
In Periscopy: EntityThe Google AI Mode feature that builds a custom interface for each query (simulations, comparison tools, interactive dashboards) instead of returning a page with a fixed template. Google made it free for everyone from summer 2026.
Generative Engine Optimization
The technical and editorial discipline that makes a brand quotable by generative AI engines (ChatGPT, Claude, Gemini, Perplexity). It covers structured data, extractable content, external authority and technical infrastructure (llms.txt, ai.txt, robots.txt). It complements classical SEO and does not replace it.
The time frame of an action in the plan: weeks 1 to 2, 3 to 6, 7 to 12, and 90 days and beyond. Every action also carries severity and effort.
In Periscopy: Plan to proofDeclaring what you want AI to say about the brand, once, for every question on that subject. The starting point is frozen on the day you declare it.
In Periscopy: InceptionAutonomous Google agents that run in the background, monitoring the web, blogs, news and market data, and alert the user when something relevant happens. They roll out from summer 2026 on the AI Pro and Ultra plans.
The queries an assistant runs on its own before answering, splitting the question into sub-questions. They show which words it searched with and who stands in the brand's place in each one.
In Periscopy: AI searchesThe internal representation, at Google and at other platforms, that links entities (companies, people, concepts) through properties such as sameAs. Appearing in the knowledge graph as a consistent entity is a strong signal for models to accept a brand as quotable.
In Periscopy: EntityWhat AI answers when it searches the web before writing. It is the second half of every measurement, next to model memory.
In Periscopy: Engines and modesLLM Optimization
The content-structure strand inside GEO. It covers how to write, structure and mark up pages so that language models extract and quote them faithfully: well-written FAQs, paragraphs that stand on their own, figures with a source, and schema.org.
A markdown file at the root of a site describing its content in a structured, model-readable form. It is an emerging standard, in use since 2024, and not every engine reads it yet. The llms-full.txt variant carries the expanded version with the main text of the pages, in a single fetch for crawlers. Periscopy checks it as part of the site's foundation, with the fix already written when it is missing.
In Periscopy: Site healthWhat AI answers about the brand without going to the internet, from what it learned in training alone. It is read separately from live search: the contrast tells you whether the problem lies in the entity or in the presence on the web.
In Periscopy: Engines and modesQuestions and sources in the category where no brand dominates yet. Whoever arrives first with quotable content takes the answer.
In Periscopy: Open territoryEvery completed action is dated and carries the citation reading from before and after. The next measurement says whether the answer moved.
In Periscopy: Plan to proofThe standard structured-data vocabulary (Organization, Service, FAQPage, Article, BreadcrumbList and others) embedded as JSON-LD in a page. It lets engines and language models understand the entities and relationships on a site without ambiguity. It is one of the technical foundations of GEO.
In Periscopy: Site healthHow much an assistant repeats the same searches from week to week. When it is high, the work has an end in sight; when it is low, the category has no owner yet.
In Periscopy: JourneySoR
The brand's share of the category's choices, not of its mentions. An answer chooses a brand when it sets it above the others it names: a place in a ranking, a verdict against a stated need (“best for frequent flyers, X”), or the closing sentence where the model commits. Being on a list of ten is not being chosen, and share of voice counts both the same. Only the answers that choose someone count; those that list and compare without choosing are left out. In the study “A map of the companies of Portugal”, across 616 answers, the rate at which a mention became a choice ran from 96% in supermarkets to 20% in the open question about the best Portuguese brands.
In Periscopy: Chosen or mentionedThe brand's share of mentions against its competitors in generated answers, measured over a fixed set of questions. It tells you not only whether you appear, but in what relative position. In Periscopy, each answer splits one unit equally among the brands it names, and the shares are averaged over the answers that name at least one brand. Being one of three is worth more than being one of twelve.
In Periscopy: Chosen or mentionedThe row of products an assistant shows inside its answer, with title, merchant, price, rating, position and the sponsored label. Periscopy reads it item by item, the way the buyer sees it.
In Periscopy: CommerceA buyer written out in full (role, context, goals, pains, criteria, vocabulary) that goes ahead of the question and changes the measured answer. It is a hypothesis about a buyer, not the buyer.
In Periscopy: Personas