Skip to content

Playbooks

Share of answer is the percentage of engine responses, across an approved set of buyer prompts, in which a brand is named or cited. We run each prompt several times per engine through official APIs and report the mean with a 95% confidence interval — a change is only a change when the intervals separate.

See our own numbers

Methodology v1.0

How we measure.

Share of answer is a number you can check, so we publish exactly how it is produced.

  • 01

    A prompt set per vertical, approved with you

    We start from a prompt library for your vertical (SaaS, D2C, manufacturing, professional services) and tune it to the questions your buyers actually ask. You approve the set before anything is run. It is refreshed quarterly and every change is versioned, so results stay comparable.

  • 02

    Five engines, official APIs, several runs per prompt

    ChatGPT, Gemini, Perplexity, Google AI Mode and Claude, queried through official APIs only. Every prompt is run several times on every engine, because answers vary between runs and a single answer is an anecdote, not a measurement.

  • 03

    Share of answer with a confidence interval

    Share of answer is the proportion of runs in which you are named or cited. We report it with a 95% confidence interval and a versioned scoring method, so a move on the dashboard means something changed in the world, not just in the sample.

  • 04

    Two more signals: AI referrals and branded search

    Share of answer is the first number. The second is an AI-referral channel grouping in your analytics, set up on the first call. The third is branded-search lift. Together they show whether being in the answer is turning into visits and demand.

How we optimise.

On-site work changes what the engines read about you. Off-site work changes who vouches for you. Most agencies stop at the first.

On-site

  1. 01

    Answer-first pages

    Each page answers its heading in the first paragraph, names the entity, and uses sub-headings phrased as questions where that is natural. We restructure the top 20 to 50 pages first, in order of the prompts they serve, and link them into answer hubs.

  2. 02

    AI-crawler access, schema and llms.txt hygiene

    Access first: robots.txt rules that let each search and AI crawler in, and key copy in the server HTML because most AI crawlers do not run JavaScript. Then schema Google supports (Organization, Product, FAQPage, Article, Service), validated before it ships. We keep an llms.txt index current as low-cost hygiene; no study yet shows it earns citations, so we never sell it as a lever.

  3. 03

    AI-crawler access

    The robots file explicitly allows GPTBot, ClaudeBot, PerplexityBot and Google-Extended alongside the search crawlers, and key copy is rendered on the server, so nothing important exists only in client-side script. A page a crawler cannot read cannot be cited.

  4. 04

    Entity pages and consistency

    About, team and product pages that say who you are in the same words everywhere, linked to the parent organisation and to your profiles elsewhere. An engine that sees three versions of a company cites none of them.

  5. 05

    Freshness and dateModified

    A publishing cadence with dated pages, a visible last-updated line, dateModified in schema and a version number on anything methodological. Fresh posts link to the answer hubs they support, so new content lifts old pages rather than competing with them.

Off-site

  1. 01

    Directories and partner pages

    Agency and company directories, partner listings and company databases, with one consistent description, founding year and category on each. These are the sources engines check for "best X in Y" prompts, and inconsistency there costs more than absence.

  2. 02

    Review platforms

    Verified reviews from real clients on the platforms your buyers and the engines trust. We set up the invitations and the cadence. We never write, buy, script or incentivise a review.

  3. 03

    Communities, with disclosed affiliation

    Answers in the forums and groups where your buyers ask, written with real data and a declared affiliation. Community threads are a source the engines cite, which is exactly why they police manipulation.

  4. 04

    Editorial and expert quotes

    Pitches to trade and business press built on original data, plus expert quotes on AI search and on your category. Editorial citations carry weight in AI answers, so they get the most effort and the longest timeline.

  5. 05

    Video with transcripts

    Short teardowns and demonstrations on a video channel, with full transcripts published on your site so the content is readable, quotable and attributable to you rather than to the platform.

What we refuse to do.

Anything that would get you removed from an engine's sources is not a shortcut. It is a liability with a delay on it.

  • Astroturfing and fake reviews

    No invented reviews, no review farms, no community posts pretending to be customers. Every review comes from a real client and every community post says who we are.

  • Private blog networks

    No link or citation schemes built on sites that exist only to link. A citation is earned from a source a buyer would actually read, or it is not worth having.

  • Scraping personal data

    We do not scrape personal or non-business data for outreach or research. Business contact data is handled under a documented legitimate-interest assessment, with a suppression list and opt-outs honoured within 24 hours.

  • Automated sending on platforms that forbid it

    No automated LinkedIn messaging and no bulk connection tools. Our outreach engine researches and drafts; a person reads, edits and sends, inside each platform's limits.

  • Promising rankings

    We promise a method, a measurement and transparency, not a position in any engine. Where fees are linked to outcomes, the outcome is share of answer on a prompt set you approved, never a ranking.

Sources.

Every number on this site links to its source. These are the sources behind the figures we quote.

  • 68% of US Google searches now end without a click. SparkToro, Jun 2026Open source
  • −58% clicks for the top result when an AI Overview appears. Ahrefs, Feb 2026Open source
  • +42% conversion from visitors who arrive from AI answers to US retail sites. Adobe Digital Insights, Apr 2026Open source
Sources behind our Insights articles 34
  • Semrush, 2026 AI Visibility Index: 126M US prompts, Jan–Apr 2026 (Jun 2026)Open source
  • Seer Interactive, AI Overviews and CTR, 2026 update: 53 brands, Jan 2025–Feb 2026 (Apr 2026)Open source
  • Ahrefs, schema and AI citations test: 1,885 already heavily cited pages, Aug 2025–Mar 2026 (May 2026)Open source
  • Ahrefs, schema and AI citations: correlation across cited pages (May 2026)Open source
  • Ahrefs, llms.txt study: 137,210 sites, May 2026 data (Jun 2026)Open source
  • SE Ranking, llms.txt study: nearly 300,000 domains (Nov 2025); data period not statedOpen source
  • Profound, where AI citations come from: 11.84B citations, 8 engines, 16 Apr–16 Jul 2026 (Jul 2026)Open source
  • Semrush, most-cited domains in AI search: weekly snapshots 14 Jul–12 Oct 2025 (Nov 2025)Open source
  • Aggarwal et al., GEO: Generative Engine Optimization, KDD 2024Open source
  • Ahrefs, brand mentions and AI Overview visibility, correlational (May 2025)Open source
  • McKinsey, The State of AI 2026: n=1,719, 97 countries, 4 May–8 Jun 2026 (Aug 2026)Open source
  • McKinsey, The State of AI 2026: organisations with $1 billion or more in revenue vs below $1 billion (Aug 2026)Open source
  • Gartner, 2026 CIO and Technology Executive Survey (Apr 2026); data period not statedOpen source
  • Microsoft, FY26 Q4 earnings, 29 Jul 2026: registered agents, not active onesOpen source
  • Ardent Partners, State of ePayables 2025 (Jan 2026); sample, period and geography not statedOpen source
  • APQC, accounts payable cost benchmarks (Mar 2026); sample and period not statedOpen source
  • Gartner, survey of 321 customer service leaders, Oct 2025 (Dec 2025)Open source
  • Gartner prediction (Feb 2026): half of companies that cut service staff due to AI will rehire, under different job titlesOpen source
  • Gartner prediction (Jun 2025): over 40% of agentic AI projects cancelled by end of 2027Open source
  • SE Ranking, AI traffic study: 101,574 websites, 250 countries, Jan–Apr 2026 (Jun 2026)Open source
  • SE Ranking, AI traffic study: share of AI referrals by engine, Jan–Apr 2026 (Jun 2026)Open source
  • Adobe Digital Insights, AI-sourced traffic report: US retail sites, Q1 2026 and Mar 2026 (Apr 2026)Open source
  • Similarweb, AI referral traffic by industry: worldwide, Jun 2025–May 2026 (Sep 2026)Open source
  • Google, AI Mode expands to more languages and locations (Oct 2025)Open source
  • G2, 2026 AI Search Insight Report: 1,076 B2B decision makers, North America, EMEA and APAC, Mar 2026Open source
  • G2, 2026 AI Search Insight Report: 1,076 B2B decision makers, Mar 2026Open source
  • G2, Buyer Behavior Report (Jul 2026); data period and region not statedOpen source
  • Forrester, State of Business Buying 2026: nearly 18,000 buyers worldwide, fielded 2025 (Jan 2026)Open source
  • Forrester, State of Business Buying 2026: buying groups, fielded 2025 (Jan 2026)Open source
  • Gartner, survey of 646 B2B buyers, Aug–Sep 2025; geography not stated (Mar 2026)Open source
  • Gartner, survey of 645 B2B buyers, Aug–Sep 2025; geography not stated (May 2026)Open source
  • TrustRadius, Beyond the hype: 1,862 technology buyers, Jan 2026 (Jul 2026); 94% is of buyers who used AIOpen source
  • EBU and BBC, News Integrity in AI Assistants: answers collected May–Jun 2025 (Oct 2025)Open source
  • Sigzen AI, published GEO prices: 34 of 41 agencies listed and 25 tools, seen 5–6 Oct 2026 (Oct 2026)Open source

The terms we use.

Short enough to quote. Each is how we use the word in reports and contracts.

  • Share of answer
    Share of answer is the proportion of AI answers, across an agreed prompt set and several runs per engine, in which a brand is named or cited. Sigzen AI reports it with a confidence interval.
  • Citation
    A citation is a link or source reference in an AI answer that points to a specific page. Being cited is stronger than being named: the engine is showing where its claim came from.
  • Named vs cited
    Named means a brand appears in the answer text. Cited means the answer links to the brand's page or a page about it. We track both, because a name without a source is easily lost.
  • AEO
    Answer engine optimisation (AEO) is on-site work that makes pages easy for AI engines to read, trust and quote: answer-first copy, structured data, crawler access and entity consistency.
  • GEO
    Generative engine optimisation (GEO) is off-site work that earns the third-party citations AI engines rely on: directories, reviews, communities, editorial and video, compliant by design.
  • Hallucination register
    A hallucination register is a log of factual errors AI engines make about a brand, such as prices, locations, products or people, with the source each error traces to and the correction made.
  • Prompt set
    A prompt set is the fixed list of buyer questions a brand is measured against, approved with the client, grouped by intent and refreshed quarterly so results are comparable over time.
  • Confidence interval
    A confidence interval is the range within which the true share of answer is likely to sit, given how many runs were made. A narrower interval means more runs and a more reliable number.

Questions buyers ask us.

  • Real, and measurable. We report share of answer — how often you're named or cited across 150 buyer prompts on five engines — with confidence intervals, before and after. The test for any provider: if the work stops at on-page SEO and reports rankings, it is rebranded SEO. GEO adds measurement across the engines and off-site citation work that SEO retainers don't cover.

Free, in two minutes. Enter your domain and we'll score it against three competitors across five engines.

No account needed. The report is emailed within 24 hours.