# Glossary > Plain definitions of the terms the Ranqo dashboard uses, from prompts, runs and engines to Share of Voice, citations, noise bands and confirmed wins. Source: https://ranqo.ai/docs/guides/glossary Every term you meet in the dashboard, in a sentence or two, with a link to the page that explains it in full. Terms are grouped by where you meet them. ## Tracking ### Prompt One question a buyer might ask an AI assistant, such as "best CRM for a small sales team". Ranqo asks your tracked prompts on every run. See [Prompts](https://ranqo.ai/docs/guides/prompts). ### Tracked, Paused and Never run The three states a prompt can be in. **Tracked** prompts are asked on every run. **Paused** prompts keep their history and sit out runs. **Never run** prompts were added but have not been tracked yet, so they have no results. ### Branded prompt A prompt that names your brand. An answer to a question about you will usually mention you, so branded prompts make visibility look higher than it is. The starter prompts are unbranded, and **Generate with AI** writes branded ones only when you pick the Comparison or Brand Research category. See [Prompts](https://ranqo.ai/docs/guides/prompts). ### Theme A group of related buyer questions built from research on your market, such as one use case or one buyer need. Themes organise your prompts, and you can generate more prompts from any theme. See [Themes](https://ranqo.ai/docs/guides/themes). ### Category The kind of question a prompt asks: Discovery, Problem/Solution, Use Case, Expert/Deep Dive, Comparison or Brand Research. Comparison and Brand Research questions usually name your brand. See [Prompts](https://ranqo.ai/docs/guides/prompts). ### Engine An AI assistant or AI search surface Ranqo asks your prompts on: ChatGPT, Claude, Perplexity, Gemini, Grok, and Google AI Overviews. Your plan decides which ones a brand can track. See [Plans and limits](https://ranqo.ai/docs/guides/plans-and-limits). ### Run One pass through a brand's tracked prompts, each asked on the engines it is set to. Runs happen every 7 days. See [Your first week](https://ranqo.ai/docs/guides/first-week). ### Snapshot The record of one run: its scores, answers and sources. The Snapshots page lists every run. See [Snapshots](https://ranqo.ai/docs/guides/snapshots). ### Reused answer When a query fails, the reading of the most recent successful answer to the same prompt on the same engine, if there is a recent one, used in its place. Snapshots mark it **imputed**. A run's own scores count reused answers; the cards on the Visibility and Competitors pages leave them out. See [When a query fails](https://ranqo.ai/docs/methodology#when-a-query-fails). ### Market Where your buyers are and what language they use: a reach (Worldwide, One country, or City or region), a target location, and a language. Ranqo asks every prompt as a buyer in that market would. See [Brand settings](https://ranqo.ai/docs/guides/brand-settings). ### Alias Another name that counts as a mention of your brand, such as an abbreviation. Aliases are matched literally. See [Brand settings](https://ranqo.ai/docs/guides/brand-settings). ## Metrics ### Visibility The share of AI answers to your tracked prompts that name your brand, with each engine weighted. One answer is one prompt asked on one engine in one run. See [Visibility](https://ranqo.ai/docs/methodology/visibility). ### Share of Voice Of every brand mention AI made in the answers, the share that was yours. See [Share of Voice](https://ranqo.ai/docs/methodology/share-of-voice). ### Average position Your average rank in an answer when AI mentions you. Lower is better. See [Average position](https://ranqo.ai/docs/methodology/average-position). ### Sentiment How positively AI talks about you, scored 0 to 100. Neutral mentions count half, so 50 is neutral. See [Sentiment](https://ranqo.ai/docs/methodology/sentiment). ### Difficulty How crowded a prompt's answers are with competitors, from None to Extreme. It is measured again from every run's answers. See [Difficulty and volume](https://ranqo.ai/docs/methodology/difficulty-and-volume). ### Volume An estimate, made by the model that writes a prompt, of how often people ask it, on a scale from very low to very high. The hover labels say "search volume", but no search engine's counts are used. See [Difficulty and volume](https://ranqo.ai/docs/methodology/difficulty-and-volume). ### Change since the previous run Change figures compare different things. The Visibility card in the Action Center compares the latest run with the one before it. The cards on the Visibility page compare the date range you picked with the range before it. The "vs last week" chip on Home compares the latest run with the most recent run that finished more than a week ago. There is nothing to compare until a brand has two runs. While the range before yours holds no run, the Visibility card on the Visibility page reads **Stable**. See [One point per run](https://ranqo.ai/docs/methodology/one-point-per-run). ### Carried value When the date range you pick contains no run, pages show the latest completed run's values instead of blanks, with an **As of** note giving its date (on the Visibility page, in the trend chart). A carried value has no change arrow. See [Carried values](https://ranqo.ai/docs/methodology/carried-values). ## Competitors ### Competitor A brand AI names in the answers to your prompts. Ranqo detects competitors from the answers; the ones you add to your watchlist are your tracked competitors. See [Competitors](https://ranqo.ai/docs/guides/competitors). ### Qualified competitor A competitor that counts toward Share of Voice: one you track, one AI named more than once, or one reported with a valid website. One-off names do not count. See [Share of Voice](https://ranqo.ai/docs/methodology/share-of-voice). ### Competitor identity Ranqo decides once, per brand, which spellings of a competitor's name are the same company, so a name does not split or merge differently from run to run. See [Competitor identity](https://ranqo.ai/docs/methodology/competitor-identity). ## Sources ### Citation A web page an engine gave as a source for its answer. Grok runs without live web search, so it returns no citations of its own; only links written into its answer text are recorded. See [Citations](https://ranqo.ai/docs/methodology/citations). ### Source A website AI cited. The Sources page lists sources by domain and by page. See [Sources](https://ranqo.ai/docs/guides/sources). ### % Answers The share of AI answers that cite a domain or page. One answer usually cites several, so the column does not add up to 100%. See [Citations](https://ranqo.ai/docs/methodology/citations). ### % Citations A domain's share of every citation AI made. Across all domains the column adds up to 100%. See [Citations](https://ranqo.ai/docs/methodology/citations). ### Source type How a cited site relates to you: You, Competitor, Company sites, Editorial, Forum, Review, Social, Video, Reference, Institutional or Other. See [Source and article types](https://ranqo.ai/docs/methodology/source-and-article-types). ### Article type The format of a cited page, such as Ranked List, How-to Guide, Comparison or Review. See [Source and article types](https://ranqo.ai/docs/methodology/source-and-article-types). ## Acting on results ### Recommendation An action in the Action Center, with steps, an expected outcome, and the metric it should move. You work it from Suggested to In Progress to Done. See [Action Center](https://ranqo.ai/docs/guides/action-center). ### Measurement window The number of runs Ranqo waits, after you mark a recommendation Done, before reading its effect. A first reading comes after 2 runs; a verdict takes more runs, how many depending on the kind of work. See [Measurement windows](https://ranqo.ai/docs/methodology/measurement-windows). ### Noise band How large a change must be before Ranqo treats it as real rather than normal run-to-run variation: 15 points for overall visibility, 20 for one engine and 27 for one prompt category. See [Noise bands](https://ranqo.ai/docs/methodology/noise-bands). ### Lift The change in a recommendation's target metric between its baseline and the runs after you marked it Done. It is shown as a number only when it clears the noise band. See [Noise bands](https://ranqo.ai/docs/methodology/noise-bands). ### Confirmed win A prompt on one engine that did not name you when the recommendation was made, and named you in both of the two most recent runs since you marked it Done. A mention in only one run counts as provisional. See [Confirmed wins](https://ranqo.ai/docs/methodology/confirmed-wins). ### Outreach target A place AI cites in your space: a Reddit community or thread, or a publication page such as a ranked list or comparison. Ranqo ranks targets by how much AI relies on them and how often those answers leave you out. See [Outreach](https://ranqo.ai/docs/guides/outreach). ## Your site ### Page Audit dimension One of the areas a Page Audit scores a page on, such as Crawlability or Citation Potential. Each has its own score and checks. See [Page Audit](https://ranqo.ai/docs/guides/page-audit). ### Site Access verdict The one state Site Access reports for your site: **Clear**, **Worth a look**, **Blocked** or **Unknown**. It is a state, not a score. See [Site Access](https://ranqo.ai/docs/guides/site-access). ### Crawler roles Site Access groups crawlers by what they do. **Live answers** crawlers fetch pages for AI assistants and AI search. **Search** crawlers index for search engines such as Google and Bing. **Training** crawlers collect data to train models; blocking them is a licensing choice, not a blocker. See [Crawlers](https://ranqo.ai/docs/methodology/crawlers). ### RanqoBot The name Ranqo's own crawler uses when it fetches your site. You can allow or block it in robots.txt. See [ranqo.ai/bot](https://ranqo.ai/bot). ### Site Tracking A small piece of code on your server that sends Ranqo the visits from AI crawlers and from people arriving via AI assistants. See [Site Tracking](https://ranqo.ai/docs/integrations/site-tracking). ## Account ### Seat One person on your account, including you. A pending invitation holds a seat until it is accepted, revoked or expires. See [Team and seats](https://ranqo.ai/docs/guides/team). ### Brand slot One brand counted against your plan's brand limit. On an unpaid plan, such as during a free trial, a removed brand that completed an analysis keeps its slot. See [Billing and trials](https://ranqo.ai/docs/guides/billing).