What is GEO?

SEO answers "can my page show up on Google?". GEO answers a different question: "when an AI assistant writes an answer, does it mention my domain in the citations?"

Most users now ask AI before clicking blue links. If AI never cites you, the user never visits — even if you'd have ranked #1 on Google. Different game, different signals, different content tactics.

Engines like Perplexity, ChatGPT with web_search, and Claude with web_search all run a two-step pipeline: retrieve a set of candidate pages, then generate an answer that quotes a few of them. Citations are the surfaced URLs.

Each engine retrieves differently — different search providers, freshness windows, and ranking heuristics. Same query, different citations. That's why Geondex tracks every engine independently per keyword.

A citation is more than a link drop. Engines record the position in the answer (top citations get read; bottom ones get skimmed), the surrounding sentence (whether your domain anchors a key claim or a side note), and a canonical URL that may differ from the URL the engine actually fetched.

Geondex normalizes URLs (strip utm_*, lowercase host, strip trailing slash, decode IDN) so example.com/blog and https://www.example.com/blog/?utm=x collapse into one citation when counting.

Engines weigh authority, freshness, and structure differently. The bars below are indicative — Perplexity leans into authority, ChatGPT rewards crisp structure, Claude pulls fresher content harder.

Real engines mix in dozens more signals, but the relative bias matters: a page that wins on one engine can be invisible on another.

Pick a query topic, tweak the imaginary "domain authority" sliders, watch which engine cites you. Simplified model, not a real engine — but the dynamics mirror reality.

• Coverage

  N of M keywords cite you (overall, across engines).

  Why it matters · Single honest number for stakeholder updates. 1/3 = you're showing up on 33% of tracked topics.

• Citation rate

  % of runs (per engine × keyword × day) that include your domain.

  Why it matters · Lower-volatility metric than rank. Drifts up as content improves.

• Avg. rank

  Position number when cited (1 = top of answer).

  Why it matters · Engines often quote in order. Rank #1 gets read; rank #8 is skim-skipped.

• Share of voice

  Your citations / total citations across competitor pool.

  Why it matters · Even if you're cited, are you cited a lot or once? SoV separates the two.

• Top competitor

  The single domain cited most across your tracked keywords.

  Why it matters · Highest-ROI study target. Open their cited URLs, learn the format.

• Gained / Lost alerts

  Diff events between consecutive runs.

  Why it matters · Causality: change you made Tue → AI started citing you Sat = signal that the change worked.

Engine ranking heuristics are private. The levers below are not — they're the highest-signal moves that consistently move citations.

• FAQ-style H2 questions

  High impact

  Section headings that match natural-language queries get quoted verbatim. <h2>How does X work?</h2> outperforms <h2>X overview</h2>.

• JSON-LD Organization + FAQPage

  High impact

  Schema markup tells engines who you are and which paragraphs answer which question. The schema audit in Geondex scores this 0-100 per URL.

• Author + dateModified

  High impact

  Engines deprecate stale content fast. A visible byline + machine-readable dateModified (in JSON-LD) is cheap to add and signals freshness.

• Quotable paragraphs (under 60 words)

  Medium

  Engines extract span-level quotes. A 200-word paragraph rarely survives the cut. Short, fact-dense paragraphs win the citation.

• Internal linking with descriptive anchors

  Medium

  Anchor text trains the retriever. read the GEO field guide beats click here.

• Canonical URL hygiene

  Medium

  Pick one URL per topic. Avoid /page, /page/, /page?utm=… variants competing for the same citation slot.

• External brand mentions

  Slow burn

  Engines that lean on authority (Perplexity especially) rank pages from domains that are co-mentioned with the query terms across the web.

• ✗ Optimizing only for ChatGPT

  ChatGPT is one engine in three. Pages tuned to its bias often crater on Perplexity (which prefers higher-authority canonical pages).

• ✗ Tracking once and forgetting

  Citation rank drifts week-over-week as engines refresh their index. Without continuous tracking, you'll never know which content change moved the line.

• ✗ Hiding behind JavaScript-only content

  Some engine retrievers don't render JS. Body-text inside React-only routes can be invisible to retrieval. Server-render the cited content.

• ✗ Stuffing keywords like it's 2012

  Modern retrievers detect spam-bait fast. Quotable, well-attributed content outranks keyword density on every engine measured.

The point of measurement isn't the chart. It's catching the moment a content change starts working — or the moment a competitor takes your spot. Geondex draws the line.

1. Tracks your domain's citation status across engines, daily or weekly.
2. Surfaces competitors — when AI cites someone else for queries you care about, you get their exact URLs. Reverse-engineer the content.
3. Alerts when a citation is gained or lost. Was your last blog post the cause? Now you know.
4. Compares engines — ChatGPT may love you while Claude ignores you. Different content fixes different engines.
5. Schema audit — crawls your domain, checks JSON-LD, scores 0-100 per URL, flags missing Organization or stale BlogPosting types.
6. Tracks rank over time. When you start getting cited at #1 instead of #5, conversion follows.

Free tier indexes up to 5 keywords on Perplexity. No card required.