Guide

AI SEO glossary: the terms that actually matter

Some links here are affiliate links. If you buy through one, we earn a commission at no extra cost to you. It never changes where a tool places —here is how we score.

Vendors in this category use the same words to mean different things. These are the definitions we use across this site.

Measurement terms

Content score — How closely your draft’s term coverage matches the pages currently ranking. A completeness checklist, not a ranking prediction. Targets worth using: the 70s in Surfer and NeuronWriter, an A or A+ in Clearscope. More on what content scores mean.

Keyword difficulty (KD) — An estimate of how hard a term is to rank for, calculated differently by every tool and comparable across none of them. MarketMuse is the only tool that adjusts it to your domain’s actual authority.

SERP weakness — Whether the pages currently ranking are beatable — forums, Reddit threads, low-authority domains. A more actionable signal than difficulty for a new site. LowFruits is built around it.

Share of voice — Your visibility across a keyword set relative to competitors, expressed as a percentage. Increasingly used for AI mentions too.

AI search terms

GEO / AEO / LLMO — Generative engine optimization, answer engine optimization, large language model optimization. Functionally synonymous. See what GEO actually is.

Prompt tracking — Running a fixed set of questions through AI engines on a schedule and recording whether you are mentioned. The core mechanic of every tool in the GEO category.

Citation — A source an AI engine references in its answer. Being cited matters more than being mentioned, because citations carry a link and the attribution.

Conversation volume — An estimate of how often people actually ask a given prompt. Currently unique to Profound, and the difference between a vanity metric and a content input.

Crawler analytics — Which named AI bots (GPTBot, ClaudeBot, PerplexityBot) visited your pages and what they extracted. Measures the input side rather than the output. Scrunch is the specialist.

Clean API environment — Querying models through an API rather than a logged-in browser. Produces consistent, reproducible data that is not necessarily what a real user sees. Most tracking tools use it; few say so prominently.

Detection terms

AI detector — A classifier measuring statistical regularity in text. It does not detect authorship, and false positives concentrate on non-native English writing. See what detector accuracy really means.

Humanizer — Software that rewrites text to disrupt the patterns detectors read as machine-generated. It adds no evidence or specificity, which is why editing beats it.

False positive — Human writing flagged as AI. Common enough that no score should ever drive a consequence on its own — what to do if it happens to you.

Pricing terms

Credit — A metered unit that means something different in every tool. Read the conversion before comparing prices, and check whether premium settings consume multiples.

Seat — A user licence. The single largest hidden cost in this category. See what tools really cost.

White-label tier gate — Which plan unlocks branded client reports. Ranges from the entry plan to $319 for the same capability, and it sets an agency’s real minimum spend.

Engine gating — Restricting which AI engines a plan can track. Some tools cap by count regardless of tier; others sell engines as add-ons; one charges eight times more for Claude via credit weighting.

Content terms

Topical authority — Demonstrated comprehensive coverage of a subject. An observed pattern rather than a metric — how to build it.

Clustering — Grouping keywords that share ranking pages onto one URL. Done by SERP overlap, not word similarity — the method.

Cannibalisation — Two of your own pages competing for one intent. The most common self-inflicted ranking problem, and the reason to map URLs before writing.

E-E-A-T — Experience, expertise, authoritativeness, trust. Three of the four can be demonstrated by good content; experience cannot be generated — why that matters.

Frequently asked questions

What is the difference between GEO, AEO and LLMO?

Very little in practice. Generative engine optimization, answer engine optimization and large language model optimization all describe influencing whether AI assistants mention your brand. Vendors pick whichever acronym suits their positioning; treat them as synonyms.

Is a content score a ranking prediction?

No. It measures how closely your draft's term coverage matches pages that already rank. Correlation with rankings exists because coverage correlates with completeness, but no vendor claims the score predicts position, and pushing it toward 100 reliably makes writing worse.

What does a credit actually buy?

It depends entirely on the tool, which is why credits make comparison hard. One credit scans 100 words on Originality.ai, 250 on Copyleaks, one keyword on Keywords Everywhere, one SERP extraction on LowFruits, and a quarter of one engine query on Rankscale - unless the engine is Claude, which costs eight times more.

Tools mentioned