We make marketing cost less, sell more.

AI answers are already deciding who gets found. We measure where your brand stands, fix the pages, and run the campaigns. An auditable baseline you can check line by line: the whole grid run, repeated passes, 95% confidence intervals and hash-sealed receipts. One fixed USD quote. Delivered in days — every number checkable by you, against the receipt. Each statement behind our copy sits on a public claims registry with its source type and its date. The counters above are counted again from that list each time this page loads.

• Words with one fixed meaningmention and citation measured apart — method
• Denominator disciplineeach rate counts only the asks that came back with an answer we could read
• The same question, seven timesseven runs per question — why seven
• Sealed receiptsinterface, model, capture time, settings hash
• Public claims registry/en/claims — source and date per entry
• Baselines in the openthe formats we deliver are open on samples

This form sends a request, and our team starts the run. What comes back: up to three buyer questions, six places where AI answers show up, one machine-run pass — the format we deliver, on your own brand, at no cost. Sample questions for your trade show up in gray. Anything you type stays yours.

What the gap costs. If nobody reads those answers back, you lose the shortlist without ever seeing the screen that decided it, and no dashboard turns red. Every reading is sealed with its conditions and your competitors are read side by side with you in the same run, which is why a figure from us can still be checked the following year.

The second bill comes due a year later. Without a dated baseline, any improvement you make has nothing to be measured against, so a win is indistinguishable from a good week and the budget argument restarts from zero every quarter. Without that second look you risk losing a quarter to a gain that never left one engine. We read 15 engines on 17 surfaces for that reason, never one platform alone. A gain on a single engine cannot be sold to you as a trend. The same protocol covers Korean and Japanese answer surfaces, which means a US brand expanding keeps one ruler at the border.

What that costs is easy to state. The shortlist gets written inside a screen you never see, out of pages you did not pick, and the names on it are settled before your analytics logs a single visit. A rank report cannot show you that screen. Neither can a brand tracker. Somebody has to read those answers back and count them, or the step that decides who even gets considered stays the one step nobody on your team can look at.

Offline, nobody signs a lease without walking the block first, because the trade area decides the business. Online, almost nobody does. CiteAngle counts whether US buyers and AI can name and cite your brand across search results and AI answers, and who is named beside you. Then it turns that proof into the pages to fix first, the sources to earn, and matched remeasurement.

The answer usually comes back as a shortlist. So your brand and the rivals it names are measured side by side in the same answer — same question, same run. You see who owns each buyer question, engine by engine.

Each finding lands as one clear move: the question, the place it ran on, the source to earn and the page to change. We read it before and after on the same panel, each place on its own denominator — the count we divide by.

Every term above is defined on one page — what the words mean. To see how the tools line up side by side, read how to choose one.

The panel is written for the market, not lifted from a corpus. Public Public sets of AI chat logs lean hard toward English — 78% of turns in LMSYS-Chat-1M (arXiv 2309.11998) and 92.35% of ShareGPT chats (arXiv 2405.01470). Even then they mirror global chat use, not the buyers in any one market. So we write the en-US panel from scratch, in US buyer wording, for your line of work. Each market gets its own panel in its own language, built from how people there really ask — never a translation of another market's set.

Each rung reuses the same engines and the same judging rules, so the numbers still line up as you climb.

The first rung is free. Send your scope and a snapshot runs one to three of your buyer questions across six places where AI answers show up. It runs as one machine-run pass, started the same business day you send it. You see the format we deliver, on your own brand, before any money moves.

Every rate we publish carries the conditions it was measured under, sealed at the moment of measurement, so a figure in next quarter's board deck can still be checked a year from now. Competitors are read in the same answer, on the same question, in the same run — a coordinate means little without the names standing beside you. Readings we planned but could not read are printed as a count, because a denominator that quietly shrinks makes every rate look better. Then the same team goes and does the work: the pages, the facts engines repeat about you, the content, and the search, social and paid channels that feed them. That includes the video. We make the video ads and the review clips in house, with models and creators under contract. We can put local creators in front of a camera in the US, Japan, China, Vietnam, Thailand and Brazil, run the paid campaigns in those countries, and handle casting, contracts, payment and likeness rights from one desk.

A paid audit runs the whole plan — each question, each run, each place. It reports every planned observation with its status: observed, cited, failed, or indeterminate, which means too vague to call. You see the whole thing, and no observation quietly leaves the count.

The US answer layer, measured surface by surface — 17 major AI and search surfaces, one grid. A surface is one place a buyer reads an answer; an engine is the service that writes it, and one engine can fill two surfaces. One pass is a story, not a read. AI answers move day to day, so each paid audit puts every buyer question to every place on this list 7 separate times and keeps all 7 replies. The swing then becomes a number. Where we read: ChatGPT, Perplexity, Claude, Grok, Gemini, Google AI Mode, Google Search, Google AI Overviews, Google News, YouTube, Google short-form video, Bing Search, Bing Copilot, Yahoo Search, DuckDuckGo search assist, DuckDuckGo organic, and Bing News.

Spec and measurement are separate lines here, and we print both. The spec above is what a paid US audit buys. The contract spec for each paid audit is the same shape at every tier: every question on that tier’s panel asked seven times across 15 engine captures, reported across 17 surfaces with a status on every cell. The US reading we have sealed on our own brand is dated July 27, 2026 — 50 questions across 15 answer surfaces, one pass. The seven-run version of that read is the next one scheduled on the US side.

One run is a point estimate, not a number — AI answers vary run to run. That is why each paid audit asks a query seven times and prints the range next to each rate it can. The full case and the math behind the ranges are public in the method.

Japan runs on the same spec. Seventeen places, every question asked seven separate times, because a Japanese answer moves the same way an American one does. Where we read: ChatGPT, Perplexity, Claude, Grok, Gemini, Google AI Mode, Google Search, Google AI Overviews, Google News, YouTube, Google short-form video, Bing Search, Bing Copilot, DuckDuckGo search assist, Bing News, Yahoo! JAPAN Search and the Yahoo! JAPAN AI answer — a Japan-only roster, distinct from the US panel above. Spec and measurement stay on separate lines in Japan too. The Japanese reading we have sealed on our own brand is dated July 17, 2026, on a roster narrower than the spec above, so every Japanese figure we quote carries its measurement date and its scope. StatCounter put Yahoo! JAPAN at 6.23% of Japanese search in June 2026. That is a door your Japanese buyers open, and we go in through it with the listing and press work below. The pass mark for the Japanese round is registered before the run, so the line a rate has to clear is set before anyone reads a number. What comes back to you is a ranked list of moves for the Japanese plain-search side, built on those observations.

We also do the Japanese-side work the reading points at. If you run stores or facilities in Japan, we take the listing work. Yahoo! JAPAN gives owners a free tool for filing facility details, and we do that filing and keep the entry current. What people see on Yahoo! JAPAN Search and Yahoo! JAPAN Maps stays accurate. The same tool counts how often the listing was shown, and it counts the taps on your website, phone number, directions and booking button. It also separates the people who searched your name from the people who found you another way. Those counts pick the next thing we fix.

Press work runs on the same footing. We write the release in Japanese from the start, not as a translation of your English one. It goes out through a Japanese distribution network to Japanese outlets. Each outlet decides what it runs, and a release that gets picked up stays online as Japanese source material that search results and AI answers can both draw on.

Every buyer query ends in one of four states. You rank and the answer carries you. You rank and it names someone else. You are missing from both. Or the answer found you with no plain-search rank at all. Each state calls for a different move, and your own distribution comes from your run.

The full FAQ — rules, what you get, how we handle data →