Skip to main content
Leaderboards use published monthly measurements. Industry, the country questions were asked from, month and side choose a leaderboard. Two kinds of filter narrow the lists on that leaderboard without changing it: brand filters, by a brand’s home country and audience, and the source type filter. A leaderboard uses exactly one market: Global, the default, Europe or one country asked from. Brand origin is a separate filter: the country a brand was founded in, even if it later moved or was acquired. Romanian brands can appear in a Global measurement or one asked from Romania.

Questions, markets and engines

Each panel measures one industry asked from one market. Global questions use English without a country. Providers that require a country receive the United States. Question sets use the market’s default language policy. Each question is asked from that market, in its recorded language. The pages stay in English. Leaderboards measure ChatGPT, Gemini, Google AI Overviews and Google AI Mode. Each question is asked once on each engine. The Command Center can also read Claude, so a leaderboard and a tracked brand can cover different answers even though they use the same scoring rule. Each monthly run also reads Google. For up to 20 of an industry’s demand keywords, the most searched first, it takes the top 10 results and reads every page that is not a brand’s own site for every brand on the leaderboard. Each question’s keyword is a short Google search phrase for the product or category a buyer needs, adapted to the question’s language and market. Questions without a dedicated keyword use the few significant words the question is about. A keyword that names a brand on the leaderboard is left out. The Search side scores these results, with each keyword weighted by its monthly search volume and a page that could not be read kept as unknown.

Europe

An industry’s Europe leaderboard pools its leaderboards in seven countries: the United Kingdom, Germany, France, Spain, Italy, the Netherlands and Romania. Europe asks no questions of its own. It combines the answers each country’s run collected. A month has a Europe leaderboard only when all seven countries published that industry. If one country is missing, that month has none and the previous month’s stays available. If a country’s reading is withdrawn, that month’s Europe leaderboard goes with it. Every question counts equally, so the countries count equally when each asked as many questions. A month whose countries asked different numbers of questions has no Europe leaderboard. Europe ranks the brands on every country’s roster when the month’s runs started. A brand added during a run appears on its country’s leaderboard only. Europe’s leaderboard pools every answer, including those from an engine some countries did not ask. Each engine’s top three covers only the engines all seven countries asked. Europe has a Search side only when all seven country reads froze one. Otherwise Search is absent and All uses the pooled AI side alone. This rule also applies to Europe’s placements on brand pages. A Europe leaderboard ranks like any industry’s, with one place per brand from the pooled answers. The public sample applies to the pooled lists. Europe compares each month’s leaderboard with the previous month’s, both pooled from their seven countries, so it has previous_month, rank changes, movers, drop-outs and a score_series like any industry’s. A brand that left the shared roster is a not_on_roster drop-out. A month without a Europe leaderboard is a missing point in the series. If any one country’s instrument changed between two months, the series starts again at the later month. The Europe hub, country=europe without an industry, lists brands across industries with their Europe places only.

Scores and places

A brand’s Heralded Score uses the questions from industries it belongs to, under the selected filters. A brand needs at least three answers naming it to receive a score and a place. Fewer naming answers leave it unranked. Each answer contributes 0 when the brand is absent, 50 when named and 100 when recommended. The score applies the product’s current calibration when you read it. Recalibration can change a published score; it preserves the ordering of scores. Published leaderboard scores use AI answers only, including when side=all. Search has no score or band. Source and brand pages keep the Heralded Score, and placements carry their available sides. Within one industry, a place is the brand’s place among all measured brands in that industry. Brand filters keep those places. Selecting Romanian CRM brands, for example, lists the Romanian brands with their CRM places, so the first row can be in third place. A ranking among a selection of brands is what a curated list is for. Across several industries, the result is a list. Naming rates from different question sets cannot establish an overall ranking. Each brand carries its places in its own industries, each with its score on that industry’s leaderboard and the number of brands that leaderboard ranks (ranked). The list orders by the best such place, then by how many industries share that place, then by name. Filtered to brands from one country, it orders by strength first: the brand’s best score on an industry leaderboard that ranks at least five brands, which says how strongly AI favours it in its own category. It has no overall rank, leader or engine top three. Use order=medals to order brands by most first places, then second places, then third places, then best placement score and name. Each brand row carries medals.first, medals.second and medals.third. A hub counts placements from industries measured today, on boards that rank at least five brands; retired industries and thinner boards do not contribute medals or placements. An industry read counts only that industry’s placement. Tied board places each earn the medal for that place. The ordering keeps the original board ranks. Use q for a case-insensitive brand-name substring search. origin_counts counts brands by hq_country before the origin filter, with any audience and name search applied. Brands with no known origin are left out of these counts. It gives you the available origins in the selected market without paging through every brand. On an industry board, each brand row carries sides.ai and sides.search, each with its rank and score. These are the side standings for that panel and month. Search is null where the brand has no measured Search score, including boards with no Search side.

Hall of fame

GET /v1/leaderboards/hall-of-fame ranks every ranked brand in the selected market scope, including brands with no podium places. Each industry measured today contributes its latest published monthly board in each market; a retired industry’s boards stay readable but count for nothing. A board counts once it ranks at least five brands, so a podium from a handful of answers does not award medals. Global and each country count once. The pooled Europe board contributes no medals or ranking placements because it reuses the countries’ answers. The order is most first places, then second places, then third places, then best rank anywhere in scope, then best Heralded Score anywhere in scope, then name. Brands equal on every key share a place; the next place skips the tied rows. Unranked brands are absent. Each row carries brand, rank, medals with first, second and third counts, best_placements, first_places and europe_first_places. The best three placements in scope each name the industry, market, rank, score and board path. They order by rank, then score, then industry and market. First places name the industry, market and board path. Europe first places are shown separately and never count toward the ranking. Use industry to count one industry’s boards. Omit market to count Global and every country, use market=global for Global alone, or an uppercase two-letter ISO country code for that country. market=europe adds up the seven European countries’ boards. exclude_country leaves one ISO country out of market=europe, or of Global and every country when market is omitted. origin_country keeps brands whose hq_country matches that ISO country. Use limit and offset to page the table. For Romanian brands abroad, Global included, use exclude_country=RO&origin_country=RO. For Romanian brands in Romania, use market=RO&origin_country=RO.

Monthly change and history

Change compares matched question and engine cells with the previous comparable month, using the paired change method. The first comparable measurement is a first reading. From the second, a change states the move between the two months’ figures, with its sign. It is up or down only when the paired test calls the move real and agrees with that sign, and otherwise steady. History follows the comparable series. A change in the measurement’s country, pooled market definition or answer interpretation version starts a new series. A scoring-version change starts a new Search series. AI and All keep the AI scale. Engine changes leave cells unpaired, as question changes do. The number of answers asked per engine does not: months with more or fewer answers form one series, and more answers narrow the intervals. Brand identity connects the history even when a brand’s name changes. New runs with Google measurements use score version v9.1. Adding Google or changing the Search score version leaves ai and all comparable. A new Search series has no previous_month, previous_rank or rank_change; its brands begin a new score_series and its change is a first reading. Earlier months stay readable on their own pages. A month without Google measurements has no search leaderboard, while ai and all continue on the AI scale. Joining another industry preserves a brand’s earlier history. Each month uses the memberships measured at that time. Changes to questions and memberships leave new or removed question cells unpaired. On a panel, each row in brands adds previous_rank, its place in the previous published comparable month, and rank_change, previous place minus current place. A positive change means a rise. These are the panel’s places under any brand filter. If either place is unavailable, the change is null. Rank movement compares monthly places directly and does not depend on the paired figure change. The row’s state is new in its first published month on the panel’s roster, returning when it has appeared before, or unranked when it has no current place. An unranked first entry has state unranked. A comparability break restarts the trend without making an existing brand new. score_series carries each comparable month’s month, rank, score and state, oldest first, starting with the brand’s first roster appearance in that series. A point is ranked, unranked, missing when the panel has no published reading that month, or not_on_roster when the panel published without the brand. Missing and absent points have no rank or score. Unpublished readings contribute no measurements. A gap does not prevent comparison with the previous published comparable month, which the board names in previous_month. The movers module has paged risers and fallers, biggest rank changes first, with ties ordered by slug. The dropouts module lists brands ranked in the previous comparable month that have no current place. Its state distinguishes unranked from not_on_roster. Each row names the brand, its previous and current place and its rank change. Brand filters narrow brands, the movers and the drop-outs to the selected brands, with their panel values. These modules use ordinary paging without withholding. Lists spanning industries have no panel rank movement, so their evolution fields and modules are empty.

Sources and questions

Sources group citations by site and type, and count both the citations and the answers that cite the site. A site’s type comes from its latest citation in the selected measurement. Rankings exclude brand sites by default, including regional domains that match a tracked brand’s domain stem or aliases. The remaining sources rank consecutively from 1, most cited first, with domain names breaking ties. Totals count the ranked sources. On the filtered read, include_brand_sites=true includes brand sites in the ranking and total. The source type filter narrows this selection. Public reads keep page counts and source-level mention availability, but withhold cited pages and per-page brand mentions and recommendations. Within one industry, each engine has its own top three measured brands. Questions carry their winners and their language. Every brand tied for the most recommendations wins that question. A question recommending nobody has no winner. The source type filter narrows sources. No filter changes what describes the whole leaderboard: the leader, questions and their winners, each engine’s top three, the brands that sources’ pages mention and recommend, the answer population and months. Each brand row also carries its Mentioned and Recommended for every engine, which is unavailable for an engine that did not read the brand. Brands, sources, each source’s cited pages, engines and questions have separate pages. Each collection carries its total, limit and offset. An empty answer population is unavailable rather than zero. months lists every month the leaderboard’s history covers, oldest first.

What the public leaderboards show

The first 20 sources on a leaderboard, hub or brand page are public, with no withheld source ranks. Positions beyond 20 are never sent. Source summaries name only the sources sent in that response. A source detail read is available when the source is among the top 20 third-party sources on at least one Global or country leaderboard. Cited-page rows and each page’s brand mentions and recommendations stay withheld on every public read, including source detail pages. Page modules return an empty rows list, their page count and the withheld positions within the first 20. Run a Snapshot for the detail about your brand: which pages are cited, which brands they mention and where yours is missing. A brand’s questions and quotes use a stable monthly sample. Heralded withholds an entity when the SHA-256 hash of its kind, identity and measurement month falls in the bottom 70% of the hash range. Questions use their question ID; quotes use their answer ID paired with the brand ID. The same entity has the same sample decision across public reads in that month. Position 1 is always sent, and positions beyond 20 are never sent. Restricted modules carry withheld, every withheld position within the first 20 rows of the ordered list. Paging uses positions before withholding. An offset of 20 or more returns no rows while keeping the same total and withheld. A page can contain fewer rows than its limit. Brand rankings, each engine’s top three, engine lists, leaderboard questions, placements, brand history_series and a source’s brands keep ordinary paging without withholding or the position-20 cutoff. The API and MCP follow the same rules. Run a Snapshot to measure your brand’s visibility and sources in a full report. The facets read adds score_bands, the Heralded Score’s bands lowest first, and pages, the catalogue of measured panel pages, saved market hubs and curated lists. Each current panel page uses its own latest published month, so a partial monthly publication keeps panels whose latest measurements are older. Retired panels keep their published pages in the catalogue. Each entry carries its path, title, description, filters, indexed, featured_position and leader. Saved hubs and lists have titles and introductions chosen for their pages. The main hub covers Global panels; the Romania hub covers panels measured from Romania. The United Kingdom, Germany, France, Spain, Italy and the Netherlands each have a hub for the panels measured from them, and Europe has a hub for its boards. A hub is empty until a panel or board publishes. “Most visible Romanian brands in AI” selects Romanian-origin brands in the panels measured from Romania and uses the strength ordering described above. facets.origins lists sorted ISO-2 brand-origin codes for brands with a measured score in at least one current published panel. Use these codes for the origin_country filter. facets.countries lists the ISO-2 codes of countries with an active panel that has published. Europe is not a country, so it is not listed there: its leaderboards appear in pages once published. Heralded stores each panel across all engines in every published month. Market hubs, brand pages and source pages are current only. Hub reads with a month return an empty read. Beside the all page, a panel or hub across all engines also stores a page for the ai side where it has a month that read Google, and one for the search side for each month that did. Everywhere else ai reads the all page, which is the same leaderboard, and search returns an empty read. A hub or curated list reads its leaderboard under its brand filters, so a change to its definition shows on the next read. A panel read without a month uses its latest stored month. A combination with nothing published returns an empty read. A publication, a brand correction or a page reading reaches the reads once Heralded has rebuilt the leaderboards it affects. Until then you receive the previous version. Each read sees one version of its leaderboard, brand page or source page, but pages you request separately during a rebuild can come from two versions. Each filter combination has a path with industry, country asked from, brand origin and audience in that order. Empty filters disappear. For example, /leaderboards/crm/romania/ selects CRM tools measured from Romania, and /leaderboards/crm/europe/ selects Europe’s CRM tools. Brand and source pages use separate paths under /leaderboards/brands/ and /leaderboards/sources/. side has no path component: the catalogue lists the all leaderboards. Titles use the filter names and measurement month. The title always names the selected audience. Global adds no country wording, and Europe’s titles say “in Europe”. A list spanning industries says “brands by AI visibility in their industry”. Descriptions name the first three measured brands in the displayed order. Page entries include featured_position, Heralded’s curated display order, and leader, the brand the page ranks first, with its ID, name, slug and domain. Both fields are null where they do not apply, including a page spanning industries with no ranked leader. A curated list’s brand_count counts its own measured brands, and its leader is its leaderboard’s. A featured path without a published panel or saved view stays out of the catalogue. The pages catalogue lists current measured panel pages, saved hubs and curated lists. Current panel pages are indexed; dated panel pages are not. Saved hubs and curated lists carry their chosen indexing status. Panel filter combinations remain available through the filtered read but do not appear as pages. Related links show the same industry in other markets, other industries in the same market and brand-origin views.

Brand and source pages

A brand page shows its best three measured placements and a paged placement module, including unranked measurements. Its places, scores and history are the published score, all, and its Mentioned and Recommended follow it. Each placement also carries sides: the brand’s rank and score on the ai and search leaderboards of the same panel and month, with the AI standing and a null Search standing while Search has no score. sides.search is null where that month has no search leaderboard, and on a Europe placement unless all seven country reads have a Search side. sides is null on a brand page Heralded has not rebuilt yet. The engine_ranks module gives each engine’s average place across the leaderboards where it ranks the brand, the number of those leaderboards and its best place. Unranked leaderboards do not enter the average. An engine with no ranked placement has no average or best place. Where a brand is on an industry’s Europe leaderboard, its Europe placement sits beside its country placements, with the place and score from that leaderboard. The questions, highlighted descriptions, sources, history and engine places come from the country leaderboards alone, so Europe adds nothing to them. Questions it answers, the questions module, lists questions the brand wins first, then questions where it places, then questions naming competitors while leaving the brand out. Each row carries the question’s language, its leaderboard title and path, the brand’s place in the question’s standing and up to five competitors, most named first. An absent brand has a null place. Standing follows the product’s order by recommendations, then mentions. Highlighted descriptions, the highlights module, select up to four complete sentences from frozen answers, one per engine. They prefer answers recommending the brand, then the most recent reading. Sentences have 60 to 280 characters; table rows, list lines and near-duplicates are left out. Each quote carries its engine, question and leaderboard. Fewer suitable quotes leave fewer descriptions. The history_series module returns one series per placement outside Europe, with its leaderboard title and path, change state and monthly rank, score, Mentioned and Recommended values. Points carry the same rank and availability state as the placement brand’s score_series, including missing months and roster departures. Mentioned and Recommended are unavailable in those gaps. A first reading and a steady series keep the monthly change states. Brand pages are current only; each placement uses its own latest published month. For a past month’s rank and score, read that month’s panel board. measured_markets maps each industry the brand is placed in to the markets with a published board for that industry, including markets where this brand is absent. Values are uppercase two-letter country codes, global and europe. To describe how many measured countries a brand leads, compare its country placements with the country codes in this list. Global and pooled Europe are separate markets. A brand’s questions and highlighted descriptions follow the public withholding rule. Highlights refer only to questions the public read can disclose. Quote selection happens before withholding. A source page groups citations by the site’s registrable domain. It shows the source’s place in each leaderboard, Europe’s included, brands its pages mention and recommend in aggregate, and the engines citing them. Each placement carries share_of_answers, the percentage of that board’s answers citing the source, from 0 to 100. An answer citing several pages of the source counts once. A Europe place is listed without adding to the aggregate counts. The cited pages and per-page brand detail remain withheld. Its page total counts combined URLs from the first 20 cited pages per source on each panel’s all-engine reading. When a page reading is unavailable, the aggregate brand mentions and recommendations remain unavailable.

Sources hall of fame

The sources hall of fame counts citations across the latest published boards of the industries measured today, on boards that rank at least five brands. It excludes brand sites. Global and each country count once per industry. The pooled Europe board never counts. Order follows the number of AI answers citing each source, then its total citations, then its domain. Each row carries domain, rank, answers, citations and boards, the number of counted boards citing it. medals with first, second and third counts, first_places and europe_first_places remain available for compatibility. Each first place names its industry, market and board path. Use industry to select one industry. Omit market to count Global and every country, use global for Global alone, or use an uppercase two-letter country code. market=europe adds the European countries’ answer and citation totals, without counting the pooled Europe board. exclude_country leaves one ISO country out, just as on the brand index. limit and offset page the source table.

Read through the API or an assistant

The leaderboard HTTP reads need no login or API key. Only published measurements appear. Use industry, country, month, side, origin_country, audience, source_type and include_brand_sites on the filtered read. Country values are uppercase two-letter codes or europe, origin values come from facets.origins, month uses YYYY-MM, and side is all (the default), ai or search. Omit country for Global. The facets read accepts month too. Brand and source detail reads are current only and have no month parameter. Add order=medals and q for a medal-ordered brand list.
Use movers_offset to page both the risers and fallers, and dropouts_offset for drop-outs. In MCP, set pages.movers and pages.dropouts independently. Each supports its own limit and offset. limit defaults to 10 and accepts 1 to 100. Every module has a separate offset, such as brands_offset, sources_offset, source_pages_offset, engines_offset or prompts_offset. Brand and source detail collections use their names too, such as placements_offset, questions_offset, engine_ranks_offset, highlights_offset or history_series_offset. On a brand read, history_series_offset pages placement series; each series contains every comparable monthly point. Each collection returns offset, limit and total; advance its offset by its limit to read another page. Restricted modules also return withheld and stop at position 20. next_cursor is null on these collections. Responses carry meta.query and meta.data_as_of, the newest included measurement. They cache for one hour and allow a stale copy for a day while revalidating, except /v1/leaderboards/facets, the page catalogue sites look new leaderboards up in, which caches for 60 seconds and allows a stale copy for another 60 seconds while revalidating. Send the ETag back in If-None-Match to receive 304 when the response is unchanged. Published responses can change after recalibration, brand corrections or page readings. Anonymous requests share a bucket per client IP, with a default of 60 requests per minute across these routes. The site’s server sends X-Heralded-Site-Key for a separate bucket, default 3,000 per minute. The site key changes only the bucket; it grants no access to unpublished measurements. An unknown key uses the anonymous bucket. RateLimit and RateLimit-Policy describe the bucket, and a 429 problem includes its state and Retry-After. The token-authenticated MCP server offers get_leaderboards for filtered modules, facets, brand pages, source pages and the brand and source halls of fame. Its normal credential limit applies. Use resource to select leaderboard, facets, brand, source, hall_of_fame or sources_hall_of_fame, and pass slug for a brand or domain for a source. For the brands hall of fame, pass paging in brand_hall_of_fame. For the sources hall of fame, pass industry, market, exclude_country, limit and offset in hall_of_fame. Leaderboard filters go in filters; collections take their own limit and offset in pages or detail_pages.