Semantic SEO Explained: 94 Terms Decoded With One Business

Semantic SEO

Koray framework term

Semantic SEO is the practice of building a website around the meaning of a subject: the things in it, the facts about them and the questions people ask. The terms below name each piece: the things are named entities, the facts are triples, and the description of how they connect is an ontology.

Google now matches the meaning of a search with the meaning of a page. A page that only repeats a phrase gives it little to work with. Koray Tuğberk Gübür popularised the modern framework from 2020. He did not coin the phrase: a 2018 academic paper by Lopezosa, Codina and Caldera-Serrano already used it.

At Example Heating: The old approach is one page that says “boiler repair Cardiff” twenty times. The semantic approach covers what a repair customer needs: common faults, the brands Rhys works on, the towns he covers and what happens next. Each page answers one need and links to the others. That is the method behind our semantic SEO and topical authority service.

Card note: The card’s “so you earn authority” is the goal of semantic SEO, not part of its definition.

Knowledge graph

Google term

The Knowledge Graph is Google’s database of facts about people, places and things, and the links between them. Google introduced it in May 2012 with the phrase “things, not strings”, meaning real objects rather than strings of letters.

Amit Singhal’s launch post on the Google blog said it held more than 500 million objects and 3.5 billion facts. Google’s Knowledge Panel Help says the facts come from public sources, licensed data and content owners. For a business owner, this is where Google keeps what it believes about your brand, your town and your products. (In lowercase, “knowledge graph” is a general term for any such database.)

At Example Heating: Google already knows that Cardiff is a city in Wales and that Vaillant makes boilers. Whether Example Heating gets an entry of its own is Google’s decision, not Rhys’s. What Rhys controls is how clearly his website states the facts that connect his firm to those known things.

Named entity

Information retrieval term

A named entity is a real-world thing with a proper name, such as a person, a place, an organisation or a product. Software can spot it in a sentence and connect it to what is already known about it.

Named entities tie your text to the real world. “Rhys in Cardiff” contains two, a person and a city, and is far easier to place than “our engineer round here”.

At Example Heating: Rhys’s pages name Rhys, Cardiff, Penarth, Barry, Worcester Bosch, Vaillant, Ideal and the Gas Safe Register. “Boiler” is not a named entity, because it has no proper name, yet it is the central entity of the whole site. Naming towns and brands outright, instead of “local areas” or “leading makes”, gives Google something specific to recognise.

Card note: The card says Google stores facts about each named entity. That simplifies things: many named entities have no entry in Google’s Knowledge Graph.

Entity attributes

Information retrieval term

Entity attributes are the properties that describe an entity, meaning a specific thing a search engine can recognise. A boiler’s attributes include its brand, its fuel type and its heat output. Each attribute paired with its value makes one fact.

Customers compare things by their attributes. They ask which brand, what size, how long it lasts and what it costs. Pages that cover the attributes people care about answer more of their real questions. The attribute is the property (brand), the value is what fills it (Worcester Bosch), and the two together make the fact.

At Example Heating: Take one combi boiler. Its attributes include brand, fuel type, heat output, warranty length and whether it suits a home with one bathroom or two. Rhys’s installation page states these values for the models he fits. His “combi vs system boiler” guide compares the same attributes side by side, which is how a homeowner thinks about the choice anyway.

Card note: The card calls attributes “the facts”. Strictly, an attribute is only the property, and it takes its value to make a fact.

Triple

Information retrieval term

A triple is one fact written in three parts. The first part, the subject, is the thing the fact is about. The middle part, the predicate, names the relationship. The last part, the object, is the other thing or value.

A 2014 Google research paper on a project called Knowledge Vault stored facts from the web in this three-part form. Google has not published how its live systems store facts. The practical point: short, factual sentences turn into triples most easily.

At Example Heating: Three facts from Rhys’s site, written as triples:

• (Example Heating, is located in, Cardiff)
• (Example Heating, installs, Worcester Bosch boilers)
• (Rhys, owns, Example Heating)

A sentence such as “Example Heating installs Worcester Bosch boilers across Cardiff” carries a clean triple. “We sort out all sorts of heating bits” carries none.

Card note: The card’s “Google stores facts this way” is a simplification, because Google has not published how its live systems store facts. The card’s “link” is properly called the predicate.

Predicate

Information retrieval term

A predicate is the middle part of a triple: the words that say how two things are related. In “Example Heating services Vaillant boilers”, the predicate is “services”. It tells a machine how the two things connect, rather than just that they appear together.

Two sentences can share every noun and still mean different things. “Example Heating services boilers” and “Example Heating sells boilers” describe two different businesses. Only the predicate changes. Structured data is the code that labels facts for search engines. It often uses a property name as the predicate, such as brand or areaServed, rather than a verb. This is not the school grammar meaning of predicate, which is the whole verb part of a sentence.

At Example Heating: Rhys’s town pages use plain predicates. Example Heating “covers” Penarth, “is registered with” the Gas Safe Register and “fits” smart thermostats. Vague phrasing such as “we’re around Penarth too” leaves the relationship unclear to a reader and to a machine.

Ontology

Information retrieval term

An ontology is a formal description of a field: the kinds of things in it, their properties and how they relate. It states, for example, that a boiler has a fuel type and that a Gas Safe engineer services it.

An ontology goes wider than a hierarchy. It also records sideways relationships, such as “a boiler needs a gas supply”. These tell you which topics belong together and which pages should link. Wikipedia’s entry on ontology notes that people often wrongly treat it as the same thing as a taxonomy.

At Example Heating: Rhys’s field has a small ontology. A boiler heats water for radiators and taps. Air trapped in a radiator leaves it cold at the top. Sludge builds up in a system, and a power flush clears it. Landlords need gas safety certificates. Each relationship suggests a link: the power flush guide explains sludge and points to the power flushing page.

Taxonomy

Information retrieval term

A taxonomy is a way of sorting things into a hierarchy of parent and child categories, so every item has one clear place. “Heating systems, then boilers, then combi boilers” is a simple taxonomy.

A clear hierarchy helps people find pages and helps search engines see how topics relate. It also shows gaps: an empty category is a topic you have not covered. In WordPress, “taxonomy” means categories and tags. In semantic SEO, it means the hierarchy of ideas behind the site.

At Example Heating: Rhys’s taxonomy runs from home heating down to boilers, radiators and controls. Under boilers sit the types (combi, system and regular), the brands (Worcester Bosch, Vaillant and Ideal) and the jobs (installation, repair and servicing). This hierarchy shapes his menus and guides, the first step in building a topical map.

These building blocks only matter because Google learned to read meaning instead of matching letters, and the short history that follows shows how that happened.

Natural language understanding

Information retrieval term

Natural language understanding is the part of computer science that teaches machines to work out what a piece of text means. It uses the words, the grammar and the surrounding context, rather than matching the exact letters someone typed.

It is one branch of natural language processing (NLP), the wider field of software that handles human language. Working out what a sentence means is often called semantic analysis. Your customers no longer need to type the “right” keyword. Your pages need to answer what people mean, not repeat what they type. Google reads meaning far better than keyword matching did, but not as well as a person. In 2019, Google’s own blog said language understanding remains an ongoing challenge.

At Example Heating: On a cold Sunday night, a family in Barry types “boiler on but radiators cold”. The word “repair” is not in it. Google still reads it as a fault, not a wish to buy a new boiler. A guide to the likely causes, pointing to Rhys’s emergency repair page, fits that search.

Card note: The card says Google reads meaning the way a person does. That overstates it.

Word embeddings

Information retrieval term

Word embeddings are lists of numbers, called vectors, that stand for words. A computer learns them by reading huge amounts of text. Words used in similar sentences end up with similar numbers, so the machine can measure how close two meanings are.

Wikipedia’s entry on the technique says a Google team led by Tomas Mikolov made it popular in 2013 with a tool called word2vec. Modern search goes further and turns whole searches and whole passages of text into numbers.

The practical result: a machine can see that two pages cover the same thing even when the wording differs. Writing naturally about your subject matters more than repeating one phrase.

At Example Heating: “Combi”, “boiler” and “hot water” appear in the same kinds of sentences across the web, so their numbers end up close. “Boiler” and “biscuits” do not. Rhys writes “the combi stopped giving hot water”. A system using embeddings can connect that sentence to a search for “boiler no hot water”.

Neural matching

Google term

Neural matching is an AI system Google added to Search in 2018. It compares the ideas behind a search with the ideas behind a page. A page can then be relevant to a search even when it lacks the exact words.

Google’s guide to its ranking systems says neural matching understands representations of concepts in queries and pages. In plain words, it matches meaning, using number lists of the kind described under word embeddings.

In practice, a page can be found for searches you never wrote down. It does not mean the page will rank. A relevant page still competes with every other relevant page on quality, links and trust.

At Example Heating: Rhys’s guide is called “Radiator cold at the top”. A tenant in Pontypridd searches “top of my radiator isn’t warm”. Few words match, yet neural matching can treat the guide as relevant, because both describe the same problem.

Synonyms and polysemy

Information retrieval term

Synonyms and polysemy are two ways that words and meanings fail to line up one to one. Synonyms are different words with the same meaning. Polysemy is one word with several related meanings. Search systems use the surrounding words to pick the intended meaning.

Google’s Search Help says a result might not contain your exact words because it has related terms instead. For an owner, that means one page per service, not one page per wording.

At Example Heating: “Combi boiler” and “combination boiler” are synonyms, so one page covers both. “Radiator” shows polysemy. It can mean the part that cools a car engine or the panel that heats a room. Both senses involve giving off heat, so they are related. Words such as “bleed” tell Google which one Rhys means.

Common mistake: treating every word with two meanings as polysemy. “Bank”, for money or a river, has two unrelated meanings, so Wikipedia’s entry on polysemy treats it as a homonym instead.

Card note: The card uses “bank” as its example of polysemy. That word is a homonym, not polysemy.

Disambiguation

Information retrieval term

Disambiguation is the process of working out which meaning of a word or name is intended, using the words around it. For search, the most important kind links a mention to the right named entity, such as a brand, a person or a place.

Koray Tuğberk Gübür calls that kind “disambiguation of the entity” in his topical authority article on Holistic SEO & Digital. The risk for a business sits in short, common words. A brand name that is also an everyday word needs clear context every time.

At Example Heating: “Ideal” is a boiler brand and also an everyday adjective. A sentence about “the Ideal Logic combi”, with “boiler” and “warranty” nearby, points to the brand. “Service” is another one. On Rhys’s site it means the annual boiler service, so his servicing page talks about flue checks and gas pressure, not customer service.

Word sequence

Information retrieval term

Word sequence is the order in which words appear in a search or a sentence. Modern language models, software trained on huge amounts of text, read each word together with its neighbours. A small word, or a change of order, can change the meaning Google takes from the text.

Google’s best-known example came in 2019 with BERT (Bidirectional Encoder Representations from Transformers), a language model that reads words in both directions. Pandu Nayak wrote on Google’s blog that, for “2019 brazil traveler to usa need a visa”, older systems missed the word “to”. They showed results about Americans going to Brazil. BERT understood a Brazilian was travelling to the US.

At Example Heating: “Hot water but no heating” and “heating but no hot water” use the same words. They describe two different faults, so Rhys’s repair guide answers each under its own heading.

Card note: The card’s car insurance example is weak, because Google probably reads both orders as the same request. The visa search makes the point better.

Co-occurrence

Information retrieval term

Co-occurrence is when two or more words or names appear near each other in text more often than chance would explain. Language researchers count these pairings, and computer systems use them to work out which words belong to the same topic.

This explains why a page written by someone who knows the trade reads differently from one built around a keyword. Real knowledge brings the right neighbouring words with it.

At Example Heating: Across the web, “boiler” keeps appearing near “pressure”, “combi”, “service” and “repair”. Rhys’s boiler repair page uses those words because a repair conversation needs them. Customers describing a fault on the phone use the same ones.

Card note: The card says using these words shows Google you know the topic. That is an inference from how the statistics work, not something Google has said.

Semantic distance

Information retrieval term

Semantic distance is a measure of how far apart two words or ideas are in meaning. A small distance means the meanings are close. Computers calculate it from sources such as word embeddings, word databases and how often words appear together.

Koray writes that “Semantic Similarity means the closeness and relevance between two words”, in his lexical semantics study on Holistic SEO & Digital. Researchers split that idea in two: similarity and relatedness. Similar things can stand in for each other; related things share a subject.

So what? Distance tells you which ideas can share a page and which need a page of their own. You make that call when you plan a topical map for your site.

At Example Heating: A combi boiler and a system boiler are similar, so one comparison guide covers both. A boiler and a radiator are related but not similar, so they get separate pages that link to each other.

Card note: The card’s advice to link near ideas comes from SEO practice, not from the definition.

Salient terms

Google term

Salient terms are the words and phrases that carry the most weight in describing what a search or a page is about. “Salient” simply means standing out. Leaked Google documents describe a score for how much each term matters as a description.

Evidence in the US Department of Justice’s case against Google describes a system called Query-Based Salient Terms. It predicts the terms a relevant page is expected to contain. Google’s Content Warehouse documents, leaked in 2024, include a salience score for terms. Leaked documents show that an idea exists, not how much weight Google gives it today.

At Example Heating: On the boiler installation page, “combi”, “flue” and “kW output” are salient, because they say what the page is about. “Great”, “reliable” and “affordable” are not. They describe nothing specific.

Card note: The card says the real terms prove you cover a topic properly, and it credits Koray. We found no Koray definition of the term.

Contextual vocabulary

SEO industry term

Contextual vocabulary is the wider set of words, names and phrases that naturally turn up when someone who knows a subject writes about it. It overlaps with salient terms but goes further, covering the whole working language of a topic.

A page missing those words can read as shallow to people and to search systems.

We could not find “contextual vocabulary” in Koray’s published articles. His verified terms nearby are topical coverage and contextual anchor text. We treat this one as a plain-English label from SEO practice.

At Example Heating: A real boiler repair page mentions the flue (the pipe that carries fumes outside). It also names the heat exchanger, the condensate pipe and the pressure gauge. Rhys uses these words because his engineers handle these parts every day. A repair page that names none of them reads as if nobody involved has opened a boiler.

Card note: The card credits Koray with this term, but we could not find it in his published articles.

Entity salience

Information retrieval term

Entity salience is a score for how central a named entity, such as a brand, product or place, is to a whole page. A high score means that thing is the page’s main subject. A low score means it is mentioned in passing.

Google’s Cloud Natural Language tool, a paid product for developers, scores salience from 0 to 1. Microsoft researchers were scoring it in 2013, before that tool existed. The Cloud score is a handy check, but Google has not confirmed that Search uses it to rank pages.

The lesson is plain: make each page clearly about one main thing. Our on-page SEO service checks this page by page.

At Example Heating: The boiler servicing page names the boiler in its title, first sentence and headings. It then explains what a service checks. A winter article that mentions boilers once, between blankets and draught excluders, gives the boiler low salience.

Source context

Koray framework term

Source context is the identity and purpose of a website as a source: what the business is, who it serves and how it earns its money. It decides which topics fit the site and how it should write about them.

Koray Tuğberk Gübür puts it simply on Medium: “The source context defines what the website represents.” He argues that Google uses it to classify the whole site. A boiler manufacturer and a local installer both write about boilers, but for different reasons and readers.

At Example Heating: Rhys’s source context fits in one line: a Gas Safe registered Cardiff heating firm that earns from installing, repairing and servicing boilers for homeowners and landlords. That line already rules topics in and out. A guide to bleeding radiators fits. A review of electric cars does not, however popular the subject.

Card note: The card says Google reads source context “before anything else”. We found no source for that, so treat it as the first thing you decide, not the first thing Google reads.

Central entity

Koray framework term

Central entity is the main thing a website is about. It is an entity, meaning a specific thing Google can recognise, and it appears on every page and in every section. Every service page and every guide comes back to it.

A clear central entity holds a site together. Each new page adds to what the site says about one thing, instead of spreading thin across many. A popular myth says the central entity must be your brand. Koray means the subject the whole site keeps returning to, not the business name.

At Example Heating: The central entity is the boiler, along with the heating system it runs. Installation, repair, servicing, power flushing and landlord certificates are all jobs done to or around a boiler. The guides follow the same thread: why a boiler loses pressure, how long a boiler lasts, combi or system. The name “Example Heating” matters too, but it is the brand entity, not the central one.

Central search intent

Koray framework term

Central search intent is the main action a website helps people take around its central entity. Glossaries of Koray’s framework describe it as the result of joining the source context with the central entity. Every topic on the site has to fit it.

It gives you a test for every page idea: a topic either helps people take that action or it does not. Koray frames the result as an action, a verb, not a label. It also differs from search intent, the goal behind one search covered in Part 4, because it governs the whole site.

At Example Heating: Rhys’s central search intent is to choose, fix and look after a boiler in Cardiff. “Combi vs system boiler” helps people choose. “Why is my boiler losing pressure” helps them fix. “How often should a boiler be serviced” helps them look after it. A post about garden decking helps with none of these, so it never gets written.

Card note: The card puts it as “why you exist joined with what you talk about”. That matches, though Koray frames the result as an action rather than a label.

Contextual domain

Koray framework term

Contextual domain is a subject area defined by three things: a concept, a topic and a context qualifier. The qualifier is the angle, audience or place that frames the topic. The same topic seen from a different context forms a different contextual domain.

Koray’s own definition is short: “A contextual domain consists of a concept, topic, and context qualifier” (topicalauthority.digital). For a business, the qualifier is the useful part, because it decides who you write for and from what angle.

At Example Heating: Home heating is the broad subject. Home heating for Cardiff homeowners and landlords is closer to Rhys’s contextual domain, because it adds an audience and a place. The qualifier changes the pages. A landlord wants gas safety certificates and their duties to tenants. A homeowner wants running costs and advice on when to replace an old boiler.

Card note: The card simplifies this to “the whole subject area your site is built to own”. That description is closer to a different Koray term, knowledge domain, meaning the wider set of concepts and topics in a field.

Topical border

Koray framework term

Topical border is the edge of the subject a website should cover. It separates the topics that belong to its source context and main subject from the topics that do not. Staying inside it keeps the site focused.

Each page outside the border spreads the site’s attention without adding to what it is known for. Koray links topical borders to nouns, meaning the things a site covers. The edge of one page’s angle is a separate idea, which he calls a contextual border. A clear border also saves money, because you stop paying for pages your customers never needed.

At Example Heating: Rhys’s border is home heating and hot water. Inside sit boilers, radiators, controls, servicing, gas safety and landlord duties. Outside sit kitchen fitting, roofing and electrical rewires, even when customers ask about them on a visit. Respecting that line is part of how topical authority builds over time.

Connected entities

SEO industry term

Connected entities are the people, places, organisations and ideas linked to an entity through stated relationships. Examples are who owns it, where it is and what field it works in. Together they help define what that entity is.

A business is understood partly through what it is connected to. The idea rests on the relationships between things that Google described when it launched its Knowledge Graph. Koray writes about “entity connections”, but connected entities is general practice vocabulary, not a Google term. For an owner, the lesson is to state each relationship plainly and the same way every time.

At Example Heating: Example Heating’s connected entities include Rhys (owner) and Cardiff (base). Penarth, Barry, Caerphilly and Pontypridd are the areas it serves. Its engineers are on the Gas Safe Register, and it fits Worcester Bosch, Vaillant and Ideal boilers. Each one appears on the site in a plain sentence. Together they tell Google this is a Cardiff heating firm, not a boiler maker or a national price comparison site.

Brand entity

SEO industry term

A brand entity is a business that a search engine recognises as one distinct real-world thing. Facts such as its name, location, website and profiles are tied together. The business is then understood as a business, not just a string of words.

Brand entity is practitioner vocabulary; Google itself talks about entities in its Knowledge Graph. You cannot register for entity status or switch it on. Google’s systems decide from information across the open web. What you control is consistency: the same name and details everywhere your business appears.

At Example Heating: Rhys makes sure “Example Heating” reads the same everywhere, with the same address and phone number. That covers his website, every directory listing and his Google Business Profile, the free listing that puts a business on Google Maps. A NAP consistency check, meaning name, address and phone, catches any mismatch. That care is a routine part of local search work for trades. It helps Google treat the firm as one business rather than several similar names.

Author entity

SEO industry term

An author entity is a named person that search engines can identify as the creator of content and connect to facts about their experience and expertise. It answers a reader’s simple question: who wrote this, and why should I trust them?

Google’s guidance on helpful content asks: “Is it self-evident to your visitors who authored your content?” The same guidance describes Experience, Expertise, Authoritativeness and Trustworthiness (E-E-A-T) and says trust is the most important of the four. Google describes no author score, so treat a clear byline as evidence for readers, not a ranking switch.

At Example Heating: Rhys writes the fault guides himself. Each carries his name, his role as owner and lead engineer and his Gas Safe registration, with a link to a short bio. A tenant reading the radiator guide can see the advice comes from the engineer who would turn up. Our own author page for this guide follows the same pattern.

Card note: The card says a known author “adds trust”. Google describes no author score, so that trust comes from readers.

Entity home

SEO industry term

An entity home is the one page, controlled by the entity itself, that search engines use as the main reference for facts about a person or organisation. For a business, it is usually the About page or the homepage.

The term was named and popularised by Jason Barnard of Kalicube, who described it in Search Engine Journal in June 2021. An entity home is the identity page for the brand or the person.

At Example Heating: Example Heating’s entity home is its About page (or homepage). It states who Rhys is, that the engineers are Gas Safe registered, the area covered around Cardiff and the services offered. It links to the firm’s profiles, and they link back. Rhys’s own bio page does the same job for him as a person.

Card note: The card credits Koray, but the term is Jason Barnard’s. Its advice, “give each service its own definitive page”, is sound site planning, but it is a different idea.

With its identity and its border settled, the business can turn to how its customers actually search, which is what Part 4 covers.

Search intent

Information retrieval term

Search intent is the goal behind a search: what the person wants to learn, find or do when they type it. The same boiler can sit behind very different goals, from fixing a fault to buying a new one.

A page that answers the wrong goal rarely gets chosen. The idea comes from information retrieval, the study of how computers find documents, and from everyday SEO practice. A key source is Andrei Broder’s 2002 paper “A Taxonomy of Web Search”. The industry uses four types. Google’s Search Quality Rater Guidelines, the handbook for the people who check its results, use their own labels.

At Example Heating: Rhys’s customers’ searches under the industry’s four types:

• Informational: “why is my boiler losing pressure”
• Navigational: “Example Heating Cardiff”
• Commercial: “combi vs system boiler”
• Transactional: “book boiler service Cardiff”

And under Google’s rater labels:

• Know: “how long does a boiler last” (Know Simple when one short answer will do, such as “how often should a boiler be serviced”)
• Do: “book a landlord gas safety certificate”
• Website: “Example Heating website”
• Visit-in-person: a place to go, such as “heating showroom Cardiff” (most of Rhys’s customers want him to come to them)

Our guide to keyword intent types has more examples.

Card note: The card’s footer credits Koray, but the idea is older.

Query semantics

Information retrieval term

Query semantics is the meaning a search engine reads from a search beyond its literal words. It uses word order, context and the things named in the search, so results match what the person means rather than the exact letters they typed.

People rarely type the words your page uses. Google’s visa example under word sequence shows how much one small word can change. Query semantics applies that kind of reading to the whole search. Pages written in plain, specific language are easier to match with the meaning behind a search.

At Example Heating: A customer types “boiler flashing a red light”. None of those words is “boiler repair”, yet the meaning is clear: the boiler has stopped because of a fault. Rhys’s repair guide describes that symptom in the customer’s own words. That lets it match the search, even without the exact phrase.

Query network

Koray framework term

Query network is the connected group of related searches around one topic. It is built from the topic, the contexts people search it in and the many ways they phrase it. A site needs to satisfy the group as a whole, not one phrase.

Koray Tuğberk Gübür popularised the phrase in SEO. In an OnCrawl article, he writes that search engines judge quality for a query or a whole network of queries. Ranking for one phrase while ignoring the searches around it leaves gaps a competitor can fill. The practical step is to list the whole network before you plan any pages.

At Example Heating: Around boiler servicing, the network might include “boiler service Cardiff” and “how often should a boiler be serviced”. It might also include “what happens during a boiler service” and “landlord gas safety certificate”. These are illustrations, not measured searches. Rhys’s servicing page and its supporting guides aim to cover the group together. A keyword research checklist helps make sure nothing obvious is missed.

Root query

SEO industry term

Root query is the broadest search for a topic, usually the name of the main thing on its own. Longer and more specific searches branch out from it. It overlaps with the familiar ideas of a seed keyword or a head term.

The root query tells you where your subject starts, and everything you plan should connect back to it. We found no published definition of this term from Koray, so we treat it as general SEO vocabulary.

At Example Heating: Rhys’s root query is “boiler”. From it branch “boiler repair”, “new boiler cost Cardiff”, “boiler losing pressure”, “boiler service” and many other variations. Few people book a job straight from the single word “boiler”. Most searches that end in a booking still build on it, which is why the boiler is also his central entity.

Representative query

Google term

A representative query is one search chosen to stand for a whole group of similar searches that share the same meaning and intent. One page can then serve the whole group, instead of one page for each wording.

The phrase comes from a Google patent on grouping similar searches (US 8,775,409, dating from 2009), which describes choosing a representative query for each group. Koray’s framework later adopted it. For a business owner, it stops a common waste: separate pages for searches that mean the same thing.

At Example Heating: The search “new boiler cost Cardiff” is the representative query for a group of similar searches. The group includes “how much is a new boiler in Cardiff” and “boiler replacement price Cardiff”. All three want the same answer, so Rhys answers them in one guide, which links to his installation page. Three near-identical pages would only compete with each other.

Query path

Koray framework term

Query path is the sequence of related searches a person makes, each one following from the last. It shows how a need develops, which helps search engines predict the next step and helps you put your pages in a sensible order.

Koray puts it plainly on Holistic SEO: “A sequential query can have a query path”. The idea comes from research on search sessions and Google patents; Koray popularised the label in SEO. A query path does not have to end in a purchase.

At Example Heating: One winter evening, a homeowner sees the pressure gauge drop. They search “boiler losing pressure”, then “why does my boiler keep losing pressure”. Both land on a helpful guide in Rhys’s outer section. It explains common causes and when to call an engineer. Next come “boiler repair near me” and “boiler repair Cardiff price”, answered by his core repair page. Step five is the booking with Example Heating.

Follow-up query

Information retrieval term

A follow-up query is the next search or question a person asks after their first one, in the same session or conversation. It narrows, extends or acts on the first need, such as moving from a problem to a price.

Follow-up queries show what people want next. Google’s May 2025 post on AI Mode, its chat-style search, describes going deeper “through follow-up questions”. A fault guide and a priced service page serve different goals, so they usually work better as separate pages joined by a clear link.

At Example Heating: After “combi vs system boiler”, a common next question is who can fit one and what it costs. Rhys does not squeeze installation prices into the comparison guide. He ends it with a short section on choosing an installer and a link to his boiler installation page, which answers the follow-up properly.

Card note: The card’s advice to “answer the whole chain on one page” is too simple. Pages with different goals usually work better apart, joined by a link.

Head and long tail

SEO industry term

Head and long tail describe two kinds of search. Head terms are short, broad searches that many people type. Long-tail terms are longer, more specific searches that fewer people type each, but there are a great many of them.

The split borrows the “long tail” idea Chris Anderson popularised in Wired in October 2004. UK figures show the shape: “semantic seo” gets about 90 searches a month, “how to do semantic seo” about 10 (DataForSEO, October 2026).

At Example Heating: The single word “boiler” is Rhys’s head term. A search such as “why does my combi boiler take so long to give hot water” is long tail. Specific searches describe a real situation, so they are easier to answer exactly. His guides are written for them, because no single page can answer everything behind “boiler”.

Card note: The card adds that long-tail terms “often convert better”. We found no published study showing this.

Query rewriting

Information retrieval term

Query rewriting is the step where a search engine turns the words someone typed into one or more alternative versions. It may fix spelling, add synonyms or add context, so it can find more relevant results.

The exact words you target therefore matter less than the meaning you cover. Query rewriting is a standard information retrieval technique used by many search engines. One well-known patent on it, filed in 2006, belongs to Yahoo, not Google. Rewriting produces variants of the search, not a guess about the person.

At Example Heating: A homeowner in a hurry types “combi boiler sevice cardif”. Search engines correct the spelling, and they may also treat “combi” and “combination boiler” as the same thing, the synonyms from Part 2. Rhys does not need a page for every misspelling or synonym. One clear servicing page, written in plain words, covers the variants.

Card note: The card says Google is “guessing intent”. That is loose, because rewriting produces variants of the search.

Query fan-out

Google term

Query fan-out is the technique Google’s AI search features use to split one question into smaller subtopics. They run several related searches across those subtopics, then combine what they find into one answer.

Google announced it for AI Mode at Google I/O on 20 May 2025, in a blog post by Elizabeth Reid. Google Search Central says its AI features, including AI Overviews, may use it, “issuing multiple related searches across subtopics and data sources”. So a page can be drawn into an answer for covering one subtopic well.

At Example Heating: Someone asks “is it worth replacing a 15-year-old boiler?”. The answer might draw on separate searches about boiler lifespan, new boiler cost in Cardiff and efficiency. Nobody outside Google can see which searches actually run, so this is an illustration. Rhys’s guides on lifespan and cost each cover one subtopic. Our guide to query fan-out for SEO goes further.

People also ask

Google term

People also ask is the Google results feature that shows a group of questions related to the search. Google’s own documentation calls it the related questions group. Each question opens to show a short answer and the page it came from.

The box shows the questions real searchers ask, which makes it a useful research source. On google.co.uk, the box for “semantic seo” includes “Can you give me an example of semantic search?” (checked 29 October 2026). Google publishes no rules for which pages it picks.

At Example Heating: A search about combi boilers might bring up “What is the downside of a combi boiler?” or “Is a combi boiler better than a system boiler?” These show the format, not the live results. Rhys answers questions like these in his guides, with the short answer first under a matching heading.

Card note: The card says to “answer them to win more space”. That is a tactic, not a guarantee.

Related searches

Google term

Related searches is the Google results feature that lists other searches people have made that relate to the current one. It helps searchers widen or narrow what they are looking for, and it often appears near the foot of the results.

Related searches are searches, not questions. Google’s documentation does not fix them to one position on the page either. They are a good source of topic ideas, but a list of related searches is not a topical map.

At Example Heating: A search for “boiler service Cardiff” might show related searches. Examples could be “boiler service cost”, “gas safety certificate Cardiff” or “boiler service near me”. These illustrate the format rather than report the live page. Each one hints at a page or section Rhys may need, and each still has to pass his topical border check before he writes anything.

Card note: The card describes it as “what else people ask”, but that describes People also ask.

Query deserves freshness (QDF)

Google term

Query deserves freshness (QDF) is Google’s idea that some searches should show newer pages, because people expect recent information. Searches where freshness matters less can keep showing older pages that still answer well.

Google’s Amit Singhal described the idea in a New York Times article, “Google Keeps Tweaking Its Search Engine”, in June 2007. Google now calls these its freshness systems, “designed to show fresher content for queries where it would be expected”.

At Example Heating: A search for “best boiler 2026” needs a current page. Rhys updates his guide to choosing a new boiler when models change. The answer to “how does a combi boiler work” barely changes, so that guide can stay as it is.

Card note: The card shortens the name and says to update time-sensitive content often. That is fair, but only for searches that need it.

How customers search tells you which pages you need, and Part 5 shows how to plan all of them as one topical map.

Topical map

Koray framework term

A topical map is the planned structure of every topic, thing and connection a website needs to cover, drawn up before any page is written. It starts from the business itself, its source context and central entity, not from a keyword list.

Koray Tuğberk Gübür made the phrase common in SEO. He wrote on Medium in June 2024 that “creating a topical map is not merely listing keywords or entities”. Without a map, a site grows one random post at a time, and posts end up competing for the same search.

At Example Heating: Boiler installation, repair, servicing and landlord certificates become money pages. Questions such as “why is my boiler losing pressure” and “combi vs system boiler” become guides that point to them. Every row is in the full map near the end of this guide, and our step-by-step topical map guide shows how to draw your own.

Common mistake: calling an AI tool’s list of 20 subtopics a topical map. Koray rejected that shortcut on Medium in February 2024.

Core section

Koray framework term

The core section is the part of a topical map built around what the business sells. It holds the commercial pages, meaning pages for people ready to hire or buy, for the most important sides of the central entity.

These pages bring in the enquiries, so they get the strongest place in the site. Koray describes the core section’s job as gathering ranking signals onto the pages that make money. Ranking signals are the evidence Google uses to choose a page, such as links pointing to it and the way searchers respond to it. Every other page in the map exists partly to support these.

At Example Heating: The core section is /boiler-installation/, /boiler-repair/, /boiler-servicing/, /central-heating/, /landlord-gas-safety-certificate/ and /emergency-boiler-repair/. Town pages such as /boiler-installation-penarth/ belong here too, because they sell the same job in a named place. A landlord in Barry who needs a certificate should land on a core page, not on a blog post.

Outer section

Koray framework term

The outer section is the part of a topical map that covers the smaller details and side questions about the central entity. Its pages are mostly informational, meaning they answer questions rather than sell, and each one points readers towards a core page.

Koray gives outer pages a specific job. On Medium in November 2025, he wrote that the outer section targets searches needing stronger historical data. That is the record of impressions and clicks that Google builds over time. An impression is one time a page is shown in results. That record then supports the core pages.

At Example Heating: “Radiator cold at the top” and “what does a power flush do” are outer guides. A homeowner in Caerphilly with a radiator cold at the top learns it probably needs bleeding. A link then takes them to the central heating page.

Common mistake: treating outer pages as filler. Each guide must stay inside the topical border; a post on kitchen fitting would not help Rhys.

Card note: The card says outer pages build relevance. Koray’s own description of their job is more specific.

Root document

Koray framework term

A root document is the main hub page of a topical map. It carries the broadest heading in the map and links out to the main sections below it. It is often the homepage, though a large site can have a separate hub for each map.

We have not found a definition of this term in Koray’s own writing. It appears in glossaries written by his students, so treat this as the working meaning used in his framework. The practical point is simple. One page sits at the top and makes it obvious what the site covers and where each section lives.

At Example Heating: The homepage is the root document. Its main heading names a Gas Safe registered boiler and heating firm in Cardiff. It links to every core service page and to the best guides. A visitor, or Google, can reach any part of the site within one or two clicks from there.

Node documents

Koray framework term

Node documents are the individual pages that make up a topical map, apart from the root document. Each one covers one part of the topic. It connects to the hub and to related pages through internal links, meaning links from one page of your site to another.

“Node” comes from graph theory, the maths of networks. A node is a single point, and the lines between points are called edges. Koray used this language in an OnCrawl case study in January 2022, where he described which node connects to which edge and with what information. The phrase “node document” comes mainly from his students’ glossaries, so we keep the definition plain: every page has a place and a job.

At Example Heating: The power flushing page is a node. So is the “what does a power flush do” guide. They link to each other because they share a customer: someone whose radiators stay cold at the bottom because of sludge.

Quality nodes

Koray framework term

Quality nodes are pages built to a clearly higher standard than the rest of a site and then given a prominent place in it. Koray says a quality node can change how a search engine judges the whole website, beyond that one page.

Search engines judge a site partly by its best work. In a Medium case study in March 2024, Koray advised linking every quality node from the homepage. A quality node is chosen on purpose. It is not simply the longest page on the site.

At Example Heating: “How much does a new boiler cost?” is a strong candidate. It answers the question every installation customer asks first. Rhys can explain what changes the price: the boiler type, the size of the home and moving the boiler to a new spot. The brand matters too, such as Worcester Bosch, Vaillant or Ideal. Linked from the homepage, it sets the standard for every other guide.

Semantic content network

Koray framework term

A semantic content network is a set of pages linked by meaning, so every subtopic and question about the central entity has its own home. Koray calls it the written and visual form of a semantic network, meaning a diagram of things and how they relate.

The links between related pages show Google how the topic fits together. Koray also warns that weak pages can pull the whole network down, so every page has to earn its place. Planning and building this network is the main work in our topical authority and content network projects.

At Example Heating: Every outer guide links to the core page it supports. Every core page links back to the guides that answer its customers’ questions. In practice, a reader can move from a question to the service that fixes it in one click.

Common mistake: treating it as a topic cluster with a new name. A cluster is usually built from keyword groups. A semantic content network is built from the entity and its attributes, meaning its facts and features.

Pillar page

SEO industry term

A pillar page is one broad, in-depth page covering a whole topic, which links out to more detailed pages on each subtopic. Those detailed pages, often called cluster pages, link back to it, so the group reads as one connected resource.

The term comes from HubSpot’s topic cluster model, not from Koray. It is useful vocabulary, but it sorts pages by how broad they are. Koray’s map sorts them by purpose instead: core section pages sell, and outer section pages answer questions and feed the core. HubSpot’s own help page notes that building topic clusters in its tool does not directly affect SEO.

At Example Heating: A “complete guide to home heating” pillar would link out to boilers, radiators and controls. It might attract readers, but nobody books an engineer from it. In Koray’s map, the money sits on /boiler-installation/ and /boiler-repair/, and the guides point there.

Cornerstone content

SEO industry term

Cornerstone content is the small set of a site’s best and most important pages, the ones it most wants to rank. They are kept long, useful and up to date, and related posts on the site link back to them.

The term comes from Yoast, the WordPress SEO plugin, which describes cornerstone content as your best and most important articles. Koray uses the word “cornerstones” in a different sense: for the core questions and answers of a topic, not for a set of flagship pages. Both ideas are useful. They simply mean different things.

At Example Heating: Rhys might mark the boiler installation page and the new boiler cost guide as cornerstone content in Yoast. The plugin then reminds him to link newer posts to them. That habit supports the same aim as Koray’s quality nodes: a few strong pages that the rest of the site points to.

Card note: The card’s footer credits Koray, which is not right. The term comes from Yoast.

Topic silo

SEO industry term

A topic silo is a way of arranging a site so that pages on one subject sit together, through folders, internal links or both. A strict silo also limits links between unrelated groups, so each group’s theme stays clear.

Siloing is usually credited to Bruce Clay, whose agency still describes it as grouping related pages together. It differs from Koray’s approach in one important way. A silo walls groups off. A semantic content network links related pages across sections wherever the meaning connects them. The clickable words of each link, called anchor text, describe the page it points to.

At Example Heating: The “how long does a boiler last” guide sits with the installation pages. A strict silo would stop it linking to the repair page. Yet a reader whose old boiler keeps breaking down may want a repair first, and that reader needs both pages.

Micro and macro context

Koray framework term

Macro context is the single main topic of one web page, set by its title, main heading and main content. Micro contexts are the smaller supporting topics inside that same page, and each adds a detail to the main topic.

Koray states the rule plainly. “Include a single macro context for every web page,” he wrote on OnCrawl in July 2022. In Koray’s usage both live inside one page. The site-wide ideas are the source context and the central entity.

At Example Heating: The “radiator cold at the top” guide has one macro context: radiators that stay cold at the top. Its micro contexts are the sections on bleeding a radiator and on when to call an engineer. Another section explains why a cold patch at the bottom is different, because it often points to sludge.

Card note: The card has the two the wrong way round: it calls micro context one page’s focus and macro context the whole site’s subject.

Contextual vector

Koray framework term

A contextual vector is the direction a page takes through its topic, shown by the order of its headings from the first to the last. A good one keeps every section moving towards the same purpose, so the page holds one angle from start to finish.

“Vector” here means direction. In 2022 Koray turned the idea into a working rule on OnCrawl: “Create a proper context vector from H1 to the last heading of the document.” The H1 is the page’s main heading. The vector is about heading flow on one page.

At Example Heating: The boiler servicing page has four headings, in the order shown below. Each heading moves the reader closer to booking. A heading on the history of the combi boiler would move the page away from that purpose.

Card note: The card says the angle is set by who you are and who you serve. That part describes source context, not the vector.

Contextual hierarchy

Koray framework term

Contextual hierarchy is the order of topics on a site or a page, from the main subject down to the small details. It decides which topic sits above which, and how much space each one gets. Koray also calls it conceptual hierarchy.

Readers and Google both use that order to work out what matters most. It shows up in three places: the heading levels on a page, the folder structure of web addresses, and how much coverage each topic receives. When the order is clear, a small detail is never mistaken for the main subject.

At Example Heating: Home heating sits at the top. Boilers sit below it, then boiler repair, then single faults such as low pressure. On the repair page, “Common boiler faults” is a main section and “Boiler losing pressure” is a subsection under it, never the other way round.

Contextual relevance

Information retrieval term

Contextual relevance is how well a page meets the real need behind a search, in the situation the searcher is in. It is judged by meaning, the things a page mentions and the words around them, rather than by exact word matches alone.

Relevance has been a central idea in information retrieval, the science of finding the right documents, for decades. Koray uses the phrase often, but he did not coin it. A page can contain every keyword and still miss what the person wanted.

At Example Heating: Someone searching “boiler repair Cardiff price” wants to know what a repair is likely to involve and how to book one locally. A page about the history of boiler design might use the same words and help nobody. The repair page is relevant because it answers the price question and shows how to book.

Card note: The card’s footer credits the term to Koray, which goes too far.

Contextual weight

SEO industry term

Contextual weight is how much importance a site gives one topic compared with its other topics. It shows in how deeply the topic is covered, how prominently it appears and how many internal links point to it.

We could not find “contextual weight” defined in Koray’s own writing. SEOs use the phrase informally. The idea fits inside contextual hierarchy, which Koray does define: the hierarchy sets the order, and the weight is how much coverage each level receives. Weight should follow what the business earns from.

At Example Heating: New boiler installs are Rhys’s big-ticket jobs, so installation gets the most depth. It has a core page, town pages, a cost guide and a combi vs system guide. Smart thermostat fitting matters, but it gets one section, not five pages. A site that gave thermostats more weight than boilers would blur what Example Heating actually does.

Card note: The card’s footer credits the term to Koray, which his own writing does not support.

Contextual bridge

Koray framework term

A contextual bridge is the connecting content between two different topics on a site. It is usually a sentence or a short section that joins one subject to the next. It lets the reader and Google follow the move.

A link is one form of bridge, but the bridge is the content that makes the link make sense. Koray wrote in January 2023 about adjusting the bridges between commercial and informational pages with the right anchor text. Commercial pages sell, informational pages answer questions, and anchor text is the clickable words in a link. A link with no bridging sentence around it gives the reader no reason to click.

At Example Heating: The “why is my boiler losing pressure” guide ends with a short section on when a pressure drop needs an engineer. Signs include a visible leak, a faulty valve, or pressure that falls again soon after topping up. That section is the bridge. Inside it sits a link to the repair page with the words “boiler repair in Cardiff”.

Card note: The card calls it a link. That describes only one form of bridge.

Main and supplementary content

Google term

Main content is the part of a page that directly helps it do its job. Supplementary content improves the visit without directly serving that job. Both terms come from Google’s Search Quality Rater Guidelines, the handbook Google gives the people who assess its search results.

The guidelines treat adverts as a separate third category, not as supplementary content. Koray refers to this split and organises his own pages into macro and micro context, but he did not create these terms. For a business owner the lesson is simple: lead with the main content and keep the extras from crowding it.

At Example Heating: On /boiler-repair/, the explanation of the repair service, the faults covered and how to book is the main content. The box of related guides is supplementary. Any paid advert on the page would be a third thing, and it should never push the repair details down.

Card note: The card’s graphic lists adverts among the extras, which the guidelines do not do.

Heading structure

SEO industry term

Heading structure is the order of headings on a page: one main heading, the H1, then H2 sections and H3 subsections inside them. It outlines the page so readers, screen readers and search engines can see how the parts relate.

Google’s SEO Starter Guide says headings in a sensible order help screen readers, the software that reads pages aloud. It also says using them out of order does not matter for Search, and there is no ideal number of them. So the gain is clarity and access, not a direct ranking reward. The same applies to semantic HTML, meaning tags such as article and section that label the parts of a page. We found no Google source saying they change rankings. Koray’s method weighs headings more heavily, because the heading flow carries the contextual vector. It is a routine check in on-page SEO for service pages.

At Example Heating: The “how often should a boiler be serviced” guide has one H1 asking that question. Its H2s cover the yearly rule of thumb, landlord duties and warranty conditions. A landlord reading on a phone can jump straight to the landlord section.

With the context of each page settled, Part 7 turns to what goes inside each section. It shows how to write pages that readers and Google choose over the rest.

Semantic content brief

SEO industry term

A semantic content brief is the written plan for one page, made before anyone writes it. It sets the page’s search intent, the questions it must answer and in which order, and the things and facts it must cover.

Koray’s framework relies on briefs, but we did not find him coining this exact phrase. A brief costs less than a rewrite. It makes sure the page fits its place in the topical map.

At Example Heating: The brief for “Why is my boiler losing pressure?” sets the intent as fixing a fault. It lists the questions in order. What pressure is normal? Why does it drop? What can a homeowner safely check, and when should they call an engineer? It names the parts to cover, such as the filling loop, and ends with the link to the repair page. Teams can learn to write briefs like this through one-to-one SEO mentoring.

Content configuration

SEO industry term

Content configuration is choosing the right format for each answer on a page. A numbered list suits steps, a table suits comparisons, and a short definition suits a “what is” question. The format follows the type of question asked.

Koray writes about “using semantic Content Formats for every semantic question type” in his topical authority article on Holistic SEO & Digital. We did not find the phrase “content configuration” in his articles, so we tag it as industry vocabulary. Google does not require a list or a table for any kind of search.

The right format saves your reader effort.

At Example Heating: “Combi vs system boiler” gets a table. It compares the space each boiler needs, how each supplies hot water and the size of home each suits. “How to bleed a radiator”, inside the cold radiator guide, gets numbered steps. “What does a power flush do?” opens with a two-sentence definition.

Information responsiveness

Koray framework term

Information responsiveness is how directly a page responds to the need behind a search. In practice, writers apply it by giving the answer straight after the question and adding the supporting detail only after that.

Koray Tuğberk Gübür named the term in his January 2023 topical map case study on Holistic SEO & Digital. He said he would explain it later, and we have not found his full published definition.

A reader with a problem decides quickly whether a page will help. Make them scroll past company history and many will leave.

At Example Heating: The losing pressure guide opens with the answer. The usual causes are a leak somewhere in the system, a faulty pressure relief valve or a problem with the expansion vessel. Then come the safe checks, the photos, and the point where a homeowner should book a repair. A slower version would open with the firm’s history and make the reader scroll down for the causes.

Card note: The card’s reading, answer the question fast, is how practitioners apply it, not Koray’s verified wording.

Answer passage

Google term

An answer passage is a short section of a page that directly answers a question. Google’s patent “Scoring candidate answer passages”, published in 2018, describes picking and scoring such passages to show as the answer to a question search.

A patent shows that Google worked out a method, not that Google uses it today in that form. Koray draws on this patent in his topical authority writing.

Write each answer passage so it makes sense on its own, without the paragraphs around it. No page is promised a quote.

At Example Heating: Under the heading “How often should a boiler be serviced?”, Rhys writes a short answer. “Once a year. The Health and Safety Executive (HSE) says gas appliances should be serviced at least once a year. Landlords must arrange a gas safety check every 12 months.” The detail on what a service includes follows below. We explain the same approach in writing service pages for answer engines.

Information density

SEO industry term

Information density is how much real, useful information a page carries for its length. A page that states clear facts in few words is dense. A page padded with filler to reach a word count is thin.

It is not a Google metric, and it is not keyword density. Google’s guidance on helpful content says Google has no preferred word count. Length should come from what the reader needs.

So what? Padding costs your reader time and costs you their trust. A customer with a broken boiler wants facts fast.

At Example Heating: A thin servicing page opens with a long paragraph about winter. A dense one lists what the service includes in its first lines: the flue check, gas pressure test, seals and safety devices. Our guide to humanising AI-written drafts shows how to cut that kind of filler.

Unique information gain

Google term

Unique information gain is the new information a page offers beyond what a reader has already seen on a topic. Google patented a way to score this, to help decide which page to show that person next.

The patent is “Contextual estimation of link information gain”, filed by Google in 2018 and granted in the US in 2022 and 2024. Its score is measured against the pages one user has already viewed, not against every page on the web. The word “unique” is an SEO addition. A patent proves Google filed the method, not that Google uses it to rank pages.

At Example Heating: Most “how long does a boiler last” pages repeat the same general advice. Rhys adds what his engineers see on real jobs. He shows photos of a corroded heat exchanger, the faults they meet in older boilers, and when a repair stops making sense. Checking what competitor pages already say shows where that gap is.

Card note: The card suggests the score is measured against every page on the web, which the patent does not say.

Consensus

Google term

Consensus is the answer that experts and trusted sources agree on for a factual question. Google’s Search Quality Rater Guidelines ask raters to check whether content matches well-established expert consensus, especially on topics that affect health, money or safety.

Raters are people who check the quality of Google’s results. Their ratings help Google test its systems; they do not rank pages directly. Koray uses the word too. “Consensus or groundedness helps search engine systems give Safe Answers”, he wrote in his entity identity study on Holistic SEO & Digital.

At Example Heating: Rhys’s landlord certificate page matches the settled fact. The Health and Safety Executive says landlords must arrange a gas safety check every 12 months. Some questions have no consensus, such as which boiler brand is best. There, Rhys gives his engineers’ view and says plainly that it is a view.

Card note: The card’s claim that Google trusts your facts more overstates it, because raters do not rank pages directly.

Comprehensiveness

SEO industry term

Comprehensiveness is how completely a page or site covers a topic. A complete page answers the questions a searcher would reasonably have. They do not need to search again somewhere else to finish the task.

Google’s own helpful content questions ask whether a page describes its topic completely, and whether readers leave feeling they need to search again. It is a general quality word. Koray’s framework uses topical coverage and contextual coverage for the idea, and he points out that the two differ.

Complete does not mean long. Google says it has no preferred word count.

At Example Heating: The new boiler cost guide covers what drives the price, fitting time, the warranty and running costs. A homeowner in Penarth can finish reading and book a survey without opening another tab. Pages of this kind, across a whole subject, are what topical authority is built from.

Card note: The card credits Koray, but this is a general quality word, not one of his terms.

Passage ranking

Google term

Passage ranking is a Google AI system that picks out individual sections, or passages, of a page. It uses them to judge how relevant the whole page is to a search. It went live for US English searches on 10 February 2021.

Google first called it passage indexing, then changed the name. Pages are still indexed, meaning stored in Google’s search index, as whole pages. The page is what ranks. The passage helps Google see that the page answers the search. When Google announced it in October 2020, it expected it to affect about 7% of searches worldwide.

At Example Heating: Rhys’s guide “How long does a boiler last?” has a section headed “Signs your boiler needs replacing”. Someone searching “signs my boiler is dying” may be shown the guide. That one section answers them, even though the page as a whole is about lifespan.

Card note: The card’s “one section can win” simplifies this, because the page is what ranks.

Featured snippet

Google term

A featured snippet is a Google result where the text answer comes first and the page link follows, reversing the normal layout. Google’s systems choose it automatically when they judge that a page’s text answers the search well.

Google’s featured snippets documentation says it does not give an exact minimum length, and you cannot mark up a page to become one. Your only control is to opt out. One way is a nosnippet tag, a line of code that tells Google not to show a snippet from the page.

No page can be built to win the box. You can make the answer passage clear and let Google decide.

At Example Heating: The power flush guide puts “What does a power flush do?” in a heading. The next sentence answers it: a power flush pumps water and a cleaning solution through the heating system to clear sludge. Google may show that as a featured snippet, or may not.

Card note: The card’s 40 to 55 words is an industry rule of thumb, not a Google rule.

Internal PageRank

SEO industry term

Internal PageRank is the share of importance that passes between pages of the same website through its internal links. PageRank itself is Google’s original method of judging how important a page is from the links pointing to it.

Google’s guide to its ranking systems says PageRank was a core system at launch and is still one today. Google publishes no internal PageRank score and names no ideal number of links per page. You control which pages your site points to most.

At Example Heating: The homepage usually earns the most links from other websites, so it has the most to pass on. A menu link from the homepage sends some of that to /boiler-installation/, the page that wins the big jobs. Guides such as “how long does a boiler last” link there too.

Contextual anchor text

SEO industry term

Contextual anchor text is the clickable wording of a link, placed inside a relevant sentence, that describes the page it points to. Readers and Google know where it leads before clicking.

Google’s link best practices say good anchor text is “descriptive, reasonably concise, and relevant”. Google flags generic wording such as “read more”. Koray’s framework connects pages “within a Semantic Content Network with contextual anchor texts”, he wrote on Holistic SEO & Digital. That is the semantic content network, a linked set of pages on one topic.

At Example Heating: The guide “how long does a boiler last” ends with: “When repairs stop making sense, see what a new boiler installation in Cardiff involves.” The words “new boiler installation in Cardiff” link to the installation page. The weak version says “find out more”.

Reasonable surfer

Google term

The reasonable surfer is a model from a Google patent that weights each link by how likely a real visitor is to click it. A prominent link in the main text gets more weight than one in a footer.

It refines the original PageRank “random surfer”, which treated every link as equally likely to be clicked. The patent is US 7,716,225, “Ranking documents based on user behavior and/or feature data”, filed in 2004 and granted in 2010. It mentions features such as footer or sidebar placement. A patent shows Google had the idea. It does not prove Google uses it today.

At Example Heating: The guide “radiator cold at the top” links to the central heating page inside the paragraph about the fix. A reader with a cold radiator is likely to click it there, far more than in a row of footer links.

Card note: The card’s word “hidden” means links in low-prominence spots, such as a footer. It does not mean hidden text, which Google’s spam policies forbid.

Orphan page

SEO industry term

An orphan page is a page on your website that no other page on the same site links to. Visitors and crawlers (programs that follow links for search engines) reach it only through a sitemap, an outside link or its address.

Google’s link best practices say every page you care about should have a link from at least one other page on your site. A sitemap, the file listing your pages for search engines, helps Google find an orphan. It gives no anchor text or context. “Orphan page” is audit vocabulary; Google’s own documents don’t use it.

At Example Heating: Rhys publishes /boiler-installation-penarth/ and forgets to link to it. No menu item or other page mentions it. Penarth homeowners rarely land on it, and Google gets no clue how it fits. The fix is two links: one from the main installation page, one from the homepage’s list of towns. A crawl in an SEO audit finds pages like this.

Backlinks

SEO industry term

Backlinks are links on other websites that point to pages on your site. Google follows them to discover pages, and its documentation says it uses links as a signal when working out how relevant a page is.

Google says “links”, not “backlinks”. Its spam policies class links made “primarily for the purpose of manipulating search rankings” as link spam. Koray describes topical authority as “a factual methodology to improve the rankings of a website without backlinks”, writing on Holistic SEO & Digital. Part 10 covers topical authority in full.

At Example Heating: Suppose a Cardiff news site covers a cold snap and links to Rhys’s guide on boilers losing pressure. That link sits on a relevant local page, in a sentence about the same problem. A batch of paid links from unrelated directories is a different thing, and the spam policies cover them.

Card note: The card calls backlinks votes; Google’s plainer word is signal.

Brand mention

SEO industry term

A brand mention is your business name appearing on another website, article or platform, with or without a link. It helps people and search systems recognise the business and connect it with its subject and its area.

A Google patent, US 8,682,892 (“Ranking search results”, granted in 2014), describes an “implied link”: a reference to a site without a clickable link. That shows Google has described the idea. It does not confirm unlinked mentions count in rankings today, and Google has not said they do.

At Example Heating: The local paper quotes Rhys in a story about landlords and gas safety checks. It names Example Heating with no link. Readers now know the name, and the firm appears next to Cardiff, landlords and gas safety. That supports its brand entity, the business as a thing Google can recognise.

Card note: The card says a mention “still counts”; the safer reading is that it helps recognition.

Structured data

Google term

Structured data is code added to a page that states its facts in a standard format, such as the business name. It usually uses Schema.org, a shared set of labels that the major search engines read.

Google’s introduction calls it “a standardized format for providing information about a page”. Done correctly, it can make a page eligible for rich results, meaning results with extra details. Eligible is not guaranteed. In August 2023 Google limited FAQ rich results to “well-known, authoritative government and health websites”. Since 7 May 2026, Google’s documentation says they no longer appear in Search at all.

At Example Heating: The boiler servicing page can carry markup naming the business, its Cardiff service area and the service. Every fact in the code must also be visible on the page. Our guide to schema markup for accountancy websites applies the same rule.

Card note: The card’s “trust” is not Google’s word: structured data describes the page.

Knowledge panel

Google term

A knowledge panel is the information box Google shows for a well-known entity, meaning a specific thing it recognises, such as a person, organisation or place. Google builds it automatically from facts gathered across the web.

Google’s Knowledge Panel Help says knowledge panels are “automatically generated”. You can’t create one with code. The entity’s official representative, such as the company itself, can claim the panel and suggest edits.

At Example Heating: A search for Worcester Bosch may show a panel about the manufacturer. A small Cardiff firm is unlikely to have one. Local customers usually see a different box for Rhys’s firm: the map listing with phone number, hours and directions. That normally comes from a Google Business Profile, which the owner sets up and manages separately. The panel and the listing both depend on accurate facts, which is the next term.

Card note: The card’s local example (address, hours, reviews) describes the map listing more than a knowledge panel.

NAP consistency

SEO industry term

NAP consistency means your business name, address and phone number (NAP) read exactly the same on your website, Google Business Profile, directories and social profiles. It is a habit from local SEO, the work of appearing in searches for your area.

Google’s Business Profile guidelines ask for your “real-world name, as used consistently on your storefront”, and an accurate address or service area. A plain reason to fix mismatches: customers ring the wrong number.

At Example Heating: The name reads “Example Heating” in the footer, the contact page, the Business Profile and every directory. Never “Example Heating Ltd” in one place and “Example Heating and Gas” in another. The phone number is identical everywhere too. The firm works in customers’ homes, so Google’s guidelines let it show a service area instead of an address. Our work on local SEO for service-area businesses follows the same rule.

Card note: The card says mismatches confuse Google; Google hasn’t published that.

E-E-A-T

Google term

E-E-A-T stands for Experience, Expertise, Authoritativeness and Trustworthiness (the guidelines often shorten the last one to Trust). These are the qualities Google’s Search Quality Rater Guidelines ask human reviewers to look for when judging whether a page can be relied on.

The guidelines say: “The most important member at the center of the E-E-A-T family is Trust.” Experience was added in December 2022. Raters judge quality; they don’t change rankings. Google says “E-E-A-T itself isn’t a specific ranking factor”, though its systems use a mix of factors that line up with it.

At Example Heating: Trust starts with the legal fact we flagged at the start of this guide. Anyone doing gas work in the UK must be Gas Safe registered, under the Gas Safety (Installation and Use) Regulations 1998. Rhys’s pages state that the firm’s engineers are registered and tell readers how to check the register. Each guide names Rhys as its author, which builds his author entity.

Card note: The card says “authority” and “signals”. Google’s word is authoritativeness.

Experience signal

SEO industry term

An experience signal is visible proof that the person behind a page has done what they write about. Their own photos, real job details, tested advice and lessons learned all count. It supports the Experience part of E-E-A-T.

Google’s Search Quality Rater Guidelines ask whether a creator has “the necessary first-hand or life experience for the topic”. “Signal” is industry shorthand; Google names no single experience ranking factor. For customers, first-hand detail separates a real tradesperson from generic advice.

At Example Heating: The guide “what does a power flush do” shows Rhys’s own before-and-after photos of a radiator flush. He explains what he checks first on a cold radiator and the mistake he sees most often.

Card note: The card also lists prices among its examples. Those count only when they are real; a page should never show made-up figures.

SERP features

SEO industry term

SERP features are the parts of a search engine results page (SERP) beyond the plain list of links. They include featured snippets, local map results, People also ask boxes, knowledge panels, videos and AI Overviews.

“SERP features” is industry vocabulary. Google’s documentation calls them visual elements, “the building blocks of the Google Search results page”. The features on a page show what kind of answer Google offers for that search. That tells you what shape your page should take, the idea behind reading the intent behind a keyword.

At Example Heating: Search “boiler repair Cardiff” and the local pack, the map of nearby firms, sits near the top. That is where Example Heating’s Business Profile competes. Search “combi vs system boiler” and a People also ask box often appears. Rhys’s guide answers those questions in short passages, the format a featured snippet draws from.

Zero-click search

SEO industry term

A zero-click search is a search that ends without the person clicking any result. The results page answered them with a snippet, a map, a phone number or an AI summary, or they searched again.

The SparkToro and Datos 2024 zero-click study, reported by Search Engine Land, looked at Google searches across every kind of query. About 58.5% in the US and 59.7% in the EU ended without a click. So a page can be seen without being visited. Being the source the results page draws on still matters.

At Example Heating: That figure covers all searches, not local trades, so it says nothing specific about Rhys’s firm. A zero-click search can still bring work. A landlord whose tenant has no heating searches “emergency boiler repair Cardiff”, sees Example Heating in the map results and taps the phone number. No website visit happens, but the call does.

These terms each cover one page or listing, so Part 10 shows how that work, repeated across a topic over time, earns topical authority.

Topical authority

Koray framework term

Topical authority is the trust a search engine gives a website across a whole subject. Its pages can then rank for many related searches, not just one keyword. It is a state a site reaches over time, not a score you can buy.

“Topical Authority is a positive ranking state,” Koray wrote on Holistic SEO & Digital in January 2023. His published formula, in an OnCrawl article from February 2021, is topical coverage multiplied by historical data. In plain words: how fully you cover the subject, times the record Google has built of how your pages perform. Both terms are explained just below. He says the idea grew from lowering Google’s cost of retrieval. Our longer guide to topical authority goes further.

At Example Heating: Topical authority would mean Google trusting the site on boiler questions in general: faults, boiler types, servicing and landlord rules. A new guide on a boiler fault would then start with that trust behind it, instead of starting from nothing.

Card note: The card divides the formula by cost of retrieval. That later version circulates widely in SEO, but his 2021 OnCrawl formula has only the two parts above.

Topical coverage

Koray framework term

Topical coverage is how completely a website covers a subject: its things, their facts and features, and the questions that connect them. It is judged by how complete and connected the information is, not by how many pages the site has.

Koray describes a topic as a graph, a web of connected things, and coverage as how well a site covers that graph (OnCrawl, February 2021). It is one half of his topical authority formula. More pages only help when each one covers a part the site was missing.

At Example Heating: Good coverage means the site answers what customers actually ask: faults, boiler types, servicing, power flushing, controls, gas safety and landlord duties. Thirty town pages that repeat the same text with a new town name add pages, not coverage.

Historical data

Google term

Historical data is the record a search engine builds about a page or a site over time. It covers how long the site has dealt with a topic, how often its pages appear in results and how searchers respond to them.

The name comes from a Google patent first filed in 2003, titled “Information retrieval based on historical data”. Koray uses it as the second half of his topical authority formula. The record includes impressions (the times a page is shown in results), clicks and signs that searchers found what they needed.

At Example Heating: Historical data cannot be rushed. Each month the guides appear for boiler searches, get clicked, and either solve the problem or send the searcher back to the results. A year of useful pages builds a record. A site launched last week has none, however good its pages are.

Card note: The card credits the term to Koray; Google named it, and Koray built on it.

Cost of retrieval

Koray framework term

Cost of retrieval is the effort a search engine spends to crawl a site, understand its pages and choose which one to show. Crawling means a search engine’s software visiting and reading pages. Koray’s advice is to keep that effort low.

In a 2023 presentation, Koray said topical authority grew from the idea of lowering this cost. His own course site names duplicate content and needless web addresses as things that raise it. A clean, fast, well organised site gives Google less work and fewer wrong choices to make.

At Example Heating: Rhys keeps the cost low with fast pages, clear headings and one page per job. Each town page carries its own local detail rather than a copy of the Cardiff page with the town name swapped. One clear boiler repair page is cheaper to process than five thin ones aimed at the same search. Speed and duplicate pages are the kind of problems technical SEO fixes deal with.

Momentum

Koray framework term

Momentum is the speed and regularity with which a site publishes new pages and updates existing ones on its topic. In Koray’s framework it multiplies the effect of topical coverage: the same coverage counts for more when it arrives at a steady pace.

Koray set this out in a Medium article in November 2025, where he defined momentum as the speed of publishing and updating. Momentum is not the short trial ranking Google gives a new page before it settles. Koray calls that trial period initial ranking and treats it as a separate idea, covered under re-ranking.

At Example Heating: Rhys picks a rhythm he can keep, such as two new guides and one refreshed page each month. Twenty guides in one week and then nothing for a year gives Google a burst, not a pattern. A steady pace also means new pages keep arriving while older ones are still being judged.

Card note: The card describes initial ranking, the short trial period for a new page, rather than momentum.

Re-ranking

Information retrieval term

Re-ranking is the stage where a search engine takes a first, quick list of results and reorders it using deeper checks. Those checks look at relevance, quality and how searchers respond, then move better pages up and weaker ones down.

Two-stage ranking, a fast first pass followed by a slower and more careful one, is standard practice in information retrieval. Koray builds on it. He treats initial ranking and re-ranking as two separate phases, and he says re-ranking relies on historical data. That is why a new page’s first position often changes. It is a starting point for testing, not a verdict.

At Example Heating: A new guide on a boiler that makes a banging noise first appears somewhere in the results for that search. Over the following weeks, Google gathers signals such as whether people find what they need or go back to try another result. The guide’s later position reflects those signals, not the first guess.

Content pruning

SEO industry term

Content pruning is reviewing a site’s weak or out-of-date pages, then improving them, merging them or removing them. A merged page usually gets a 301 redirect, a permanent forward from the old web address to the new one.

In August 2023, Search Engine Land reported a comment from Google’s Danny Sullivan. Deleting older content does not by itself make Google see a site as better, he said. At most, it may help crawling on a very large site. Google’s helpful content guidance also lists removing lots of older content to seem fresh as a warning sign. Prune for readers.

At Example Heating: Suppose Rhys has three short posts that each answer “how often should a boiler be serviced”. Merging them into one guide, with redirects from the old addresses, gives readers one clear answer. A page-by-page SEO audit finds overlaps of this kind. Deleting an older radiator guide that still helps people would gain nothing.

Card note: The card’s example says merging thin posts can lift the whole site. Google has not said that.

Content freshness

SEO industry term

Content freshness is how current a page’s information is. It is kept by updating the facts, figures, prices and guidance on the page when they change, not by changing the date at the top.

Freshness matters more for some searches than others. Google’s guide to its ranking systems says its freshness systems show newer content where searchers would expect it, which is the idea behind deserves freshness. Google’s helpful content guidance also treats changing page dates just to look fresh as a warning sign.

At Example Heating: Rhys updates a page when something real changes. That might be new guidance from the Gas Safe Register or the Health and Safety Executive (HSE). It might be a new boiler range from Worcester Bosch, Vaillant or Ideal, or a change to a grant scheme. The landlord certificate page needs the most care. The HSE says landlords must have the gas appliances they provide safety checked every year, so stale advice there could mislead a landlord.

AI Overviews

Google term

AI Overviews are the AI-written summaries Google shows at the top of some search results. They give the gist of an answer with links to supporting web pages, using a version of Gemini, Google’s AI model, customised for Search.

Google launched them to everyone in the US in May 2024. They appear only when Google’s systems judge them “additive”, meaning useful beyond the normal results. Google states: “There are no additional requirements to appear in AI Overviews or AI Mode”. A page only needs to be indexed and eligible to show a snippet. Google can also use query fan-out, splitting one question into several searches.

At Example Heating: Someone in Barry searches “radiator cold at the top”. An AI Overview explains that trapped air needs bleeding out and links to a few guides. To be eligible, Rhys’s guide must be indexed and able to show a snippet. A plain answer near the top makes it easier to use. How we approach AI Overviews covers the rest.

Card note: The card’s “pages it trusts” oversimplifies the rule.

Grounding

Google term

Grounding is the step where an AI model ties its answer to real sources it has looked up, such as search results or documents. The model then relies on more than its training. A common way to do this is retrieval-augmented generation (RAG). The system first finds relevant documents, often with semantic search, which matches meaning rather than exact words. It then writes its answer from them.

Google Cloud’s documentation describes grounding as a way to “Connect model output to verifiable data sources, reducing hallucinations”. A hallucination is a confident answer that is false. The term is used in Google’s AI products and across AI research. Grounding reduces errors; it does not remove them.

At Example Heating: Someone asks an AI assistant how often a boiler should be serviced. A grounded assistant searches, reads the pages it finds, then answers from them. Rhys’s servicing guide is easier to use as a source when its first sentence gives the answer.

Card note: The card says clear pages get pulled in. That is a sensible tactic, not a documented rule.

AI citations

SEO industry term

AI citations are the links or named sources an AI answer shows to back up what it says. Google’s AI Overviews and AI Mode, ChatGPT and other assistants attach them so readers can check the source.

Google’s own phrase is “links to supporting websites”. A citation shows your page to the reader but guarantees no click. Ahrefs has measured how many pages cited in AI Overviews also rank in the top 10. Its July 2025 study found about 76%. Its March 2026 study of 863,000 results pages found about 38%. Ranking in the top 10 does not guarantee a citation.

At Example Heating: Rhys can’t choose to be cited. He can make each page worth citing: one clear answer per question, his name on the guides, and the same firm facts everywhere. Answering the smaller questions inside a big one gives more ways to match, as our query fan-out guide explains. In practice, his guide could rank on page one and not be cited, or be cited without ranking there.

Card note: The card’s “clear, trusted pages get quoted” describes better odds, not a promise.

Answer engine optimisation (AEO)

SEO industry term

Answer engine optimisation (AEO) is shaping content and business facts so search engines and AI assistants can lift them out as a direct answer. It covers featured snippets, voice answers, AI Overviews and chatbot replies.

AEO is industry vocabulary, and so is generative engine optimisation (GEO), a newer label for similar work aimed at AI-written answers. Jason Barnard of Kalicube says he coined it in 2017; that is his own claim. Koray’s framework supplies the groundwork, starting with source context. As quoted under AI Overviews, Google sets no extra requirements for its AI features. So AEO is good SEO with clear answers and clear entities.

At Example Heating: The landlord certificate page says what the certificate covers in its first sentence. Brand facts match everywhere, and the guides cover each step of the query path. Rhys then checks which AI answers mention the firm and fills the gaps, as our AI search optimisation service does.

Card note: The card’s title says “Answer engine SEO” and its footer credits Koray, but the term is not his.

AI search rewards the same foundations as everything before it, so the next section puts them together: Example Heating’s full topical map.