Loading Scope
Measurement methodology
What your report measures, how comparisons stay fair, and where the limits are. The full detail, in plain language.
Scope measures what AI actually recommends, from dated, saved answers. Not from rankings, crawler access or one unexplained score.
Recommendation rate is the share of customer answers that actively recommend you, counted only on questions that don’t name your brand. Being recognised or mentioned in passing doesn’t count.
AI visibility counts every answer where you come up at all: recommended, preferred, mentioned, warned against or ruled out. Visibility tells you whether you’re in the conversation; recommendation rate tells you whether you’re chosen.
Competitors are measured on the same questions, assistants, market and dates as you.
A business has to keep coming up across separate answers before Scope calls it a real competitor. Scope keeps two lists: everyone AI recommends, exactly as named, and your competitors, filtered for market, size and business model. We check that second list against a hand-labelled benchmark of 24 businesses in 7 languages, where every competitor shown was checked and labelled by hand: more than 8 in 10 are real rivals.
Scope records the pages AI cites and sorts them into your site, competitor sites and other sources. Search-engine plumbing and broken links are left out.
Every business named, every excerpt quoted and every page Scope suggests you change is checked against the captured answers and the pages it actually fetched. We tested this on real audits across retail, software, services, hospitality and banking, and found no invented claim.
Scope asks as many customer questions as a category genuinely supports, rather than padding every audit to one fixed size. A bank is measured on the decisions its customers weigh, such as eligibility, fees, deposit protection and switching, not on generic discovery.
AI assistants rarely give the same answer twice.
Paid audits repeat the questions that matter most, so a result reflects a pattern rather than one lucky or unlucky response. The free check asks once, so it is a first look rather than a stable measurement.
Scope only compares audits that asked the same questions in the same market. A change counts when it shows up across questions, assistants or months, not in one lucky answer.
An audit is a dated sample of AI answers.
It cannot guarantee what AI will say next, tell you how many people ask each question, or prove that one website change caused a result. Citations show what AI leaned on, not why it chose. Monitoring is monthly, and Scope currently asks ChatGPT, Claude and Gemini only. AI Overviews and Perplexity: coming in October.
Walk through a complete report: where you stand in each market, the rivals AI recommends instead, and the fix that comes next. The same report your clients or your team will read.
Open the example reportInteractive · Fictional business · No signup needed
