Link building glossary

The visible, clickable words inside a link. Search engines have used anchor text since the earliest days of link analysis as a description of what the destination page is about, which is why it carries more weight than its length suggests. It is also the most obvious signal of manipulation: a natural profile contains a messy mixture of brand names, page titles, bare URLs and phrases like click here.

The proportions of different anchor types across a whole link profile: branded, naked URL, generic, partial match and exact match. Distribution is far more diagnostic than any single anchor, because manipulation shows up as a shape rather than as one bad link. Comparing your distribution to several competitors in the same market is the practical way to read it.

Semrush's composite domain metric, which blends link signals with estimated organic traffic and measures of spam. Because it includes traffic, it can rate a site with modest links but real visitors above one with the reverse — a genuine difference in emphasis from the purely link-based scores, and a reason to check more than one tool.

A site that accepted submitted articles from anyone, with author links attached, on the theory that free content in exchange for a link benefited both parties. The model was industrialized with spun text, discounted comprehensively, and is now graded obsolete rather than high risk: it will not hurt you, it simply does nothing at all.

Search experiences that generate an answer and cite sources rather than only returning a list of links. The selection process rewards being a clearly identified, consistently described, frequently referenced source on a subject — which overlaps with what earns links, but weights named mentions and clarity of identity more heavily than hyperlink counts.

A link on one website pointing at a page on another. From the receiving page's point of view it is an inbound link; from the linking page's point of view it is an outbound one. Backlinks are the raw material of link building, but counting them tells you very little on its own, because a single site can send thousands through one template.

Anchor text consisting of a company, product or domain name — the name itself, the domain, or a phrase built around it. Branded anchors dominate genuinely organic profiles, because when people link to you unprompted they usually refer to you by name. A profile with very few branded anchors and many commercial phrases is unusual in a way that invites scrutiny.

The independently crawled database of links maintained by a tool vendor. Every vendor crawls the web itself, so no two indexes contain the same links, and none of them contains everything. This is why two tools report different totals for the same site, and why comparing a figure from one tool with a figure from another is meaningless.

Finding links on other people's pages that point at dead URLs, then contacting the page owner to suggest a working replacement you happen to publish. The ask is legitimate on its face, which is why it is graded safe, and the work is almost entirely research. The failure mode is suggesting a replacement that does not genuinely cover what the dead page covered.

A link placed in the comment section of somebody else's post. Automated comment spam was one of the earliest link tactics at scale and prompted the invention of the nofollow attribute; comment links now carry nofollow or UGC attributes essentially everywhere. Commenting to join a conversation is worthwhile. Commenting for links is not.

Shorthand for tactics that knowingly breach search engine guidelines. Like its counterpart the term obscures more than it explains: it lumps genuinely dangerous schemes together with tactics that are merely dead, and it invites arguments about labels instead of about mechanisms, consequences and who carries the risk when it goes wrong.

Any published reference to your organization by name, linked or not. Mentions contribute to how confidently a search engine can associate your name with a subject, and they are the primary currency in systems that cite sources by name rather than by hyperlink. Tracking them is now as worthwhile as tracking links.

A structured review of everything linking to a site: where the links come from, how they were acquired, which are lost, which are risky, and what the profile's overall shape says. It is the first work to do when inheriting a site, and it should combine data from more than one index because no single index sees everything.

A link placed inside the body copy of a page, surrounded by relevant text that gives it meaning. Contextual links are generally the most valuable kind because their position implies an editorial decision — somebody writing a sentence chose to point at you. The term is also heavily used in link sales, where it usually means an insertion into an existing article.

Two pages being mentioned or linked to by the same third page. If an article about accounting software links to you and to two well-known competitors, that co-citation associates you with them in the graph, whether or not the link passes measurable ranking value. It is one of the mechanisms by which mentions without links can still carry weight.

The appearance of your brand name near particular words and topics across the web, independent of any link. Where co-citation concerns shared links, co-occurrence concerns shared language: being repeatedly written about alongside the terms that define your market helps a search engine understand what you are, and it needs no hyperlink to do so.

Majestic's measure, also on a 0 to 100 scale, of how much link volume points at a page or domain, without regard to the quality of the sources. On its own it says almost nothing, because volume is trivially easy to manufacture. Its value comes from the comparison with Trust Flow, which is what the trust ratio expresses.

An HTML element that tells search engines which URL is the preferred version of a page when several show similar content. It matters for link building because links pointing at non-canonical duplicates are consolidated toward the canonical version, and a misconfigured canonical can quietly send the value of your acquired links to the wrong page or nowhere.

The practical limit on how much of a site a search engine will crawl in a given period, determined by the site's size, its server responsiveness and how much demand there is for its content. It is a genuine constraint on very large sites and an irrelevance on small ones, despite being discussed as though every site were affected.

An ordinary link with no rel attribute restricting how search engines treat it. There is no actual dofollow value in HTML; the word exists only as the opposite of nofollow. A dofollow link is simply the default state of a hyperlink, and the industry's habit of treating it as a special category has caused more confusion than it has cleared up.

Moz's 0 to 100 predictive score for how likely a domain is to rank, based on its link profile. It is a third-party estimate produced from a third-party crawl, not a figure Google publishes or uses. It is useful for comparing prospects on the same scale and actively harmful when treated as a threshold to buy against, which is how it is usually sold.

Ahrefs' 0 to 100 score for the strength of a domain's backlink profile, calculated on their own crawl of the web. Domain Rating and Domain Authority measure similar things on similar scales from different data, so they frequently disagree about the same site. Neither is wrong; they are two independent estimates of something neither can observe directly.

Earning coverage and citations in media publications through stories journalists actually want to run: original research, expert commentary, data with a genuine angle. It is public relations practiced with an awareness of search, not link building with a press release attached. The links are a consequence of the coverage rather than the thing being negotiated.

A text file submitted to Google listing URLs or domains whose links you want ignored when your site is assessed. It is the correct tool for cleaning up after a manual action or a genuinely manipulative history, and the wrong tool for almost everything else. Disavowing links on the basis of a third-party toxicity score routinely removes value from perfectly good links.

Removal of a page or an entire site from a search index, so that it cannot appear in results at all. For link building the relevant case is a network or a low-quality publisher being deindexed, which instantly removes any value the links on it were providing — and is one of the ways a purchased link portfolio evaporates without notice.

A link given because somebody decided independently that a page was worth citing, with no money, trade or request involved. It is the standard every other kind of link is measured against, and the reason links function as a ranking signal at all. In practice the line is blurry: a link earned by a good outreach email is still an editorial decision.

Anchor text identical to the keyword phrase the destination page is trying to rank for. It was extraordinarily effective early on, was industrialized, and became one of the clearest signals of manipulation as a result. Occasional exact match anchors occur naturally. A pattern of them pointing at commercial pages does not, and is straightforward to detect.

A domain whose registration lapsed and which can be re-registered by somebody else, sometimes retaining historic links. Buying expired domains has legitimate uses — recovering a name, consolidating a brand, protecting a trademark — and one notorious one, which is rebuilding them as a network or redirecting their accumulated links at an unrelated site.

Whether the messages you send arrive in an inbox rather than a spam folder or nowhere at all. Deliverability depends on your domain's sending reputation, correct authentication, sending volume and how recipients react. Outreach programs that scale carelessly destroy it, and a damaged sending domain silently ruins every campaign that comes after.

The set of DNS records that let a receiving mail server verify a message really came from your domain, principally SPF, DKIM and DMARC. Without them, outreach is far more likely to be filtered before anyone sees it. Configuring authentication correctly is unglamorous technical work and it has more effect on outreach results than most changes to the email copy.

A distinct thing a search engine can identify and hold information about: a person, company, product, place, event or concept. Modern search reasons about entities rather than only about strings of text, which is why being consistently named, described and connected across the web now matters alongside being linked to.

Experience, expertise, authoritativeness and trustworthiness — the framework in Google's published quality rater guidelines describing what raters assess. It is not a score and not a ranking factor, but it names what the systems are attempting to approximate, and links and citations from recognized sources are among the clearest external evidence of it.

Any repeated, detectable characteristic that reveals a group of sites is operated by the same party or produced by the same process. Shared hosting, identical themes, matching registrant details, the same publishing schedule, the same outbound link patterns. Finding footprints is how networks are identified, by search engines and by anyone auditing a competitor.

Majestic's index of pages crawled recently, typically covering a rolling window of the last few months. It is where you look to see what has just happened — new coverage, a campaign landing, links appearing after a launch. Because the window is short, it undercounts the total profile and should never be compared against a historic figure.

An article written for somebody else's publication, normally carrying a link to the author's site. The same two words cover two entirely different things: a genuine contributed piece to a publication that would have commissioned it anyway, and a placement bought at volume on sites that exist to sell them. The first is editorial; the second is a link scheme.

The emerging label for work aimed at being selected and cited by systems that generate answers. Much of it is ordinary good practice under a new name: be genuinely authoritative on a defined subject, be structured and unambiguous, be widely and consistently referenced. Treat confident claims about how these systems rank sources with suspicion; the field is young.

Tying a generated answer to specific retrieved sources so its claims can be attributed and checked. Where an answer is grounded, there are citations, and those citations are the visibility being competed for. Pages that state a fact plainly, with evidence and a clear author, are easier to ground against than pages that bury the point.

Google's free reporting tool for site owners, showing how the site appears in search, which queries produce impressions and clicks, indexation status, manual actions and a sample of known external links. It is the only link data that comes from Google rather than from a third-party crawl, which makes it the reference point for any audit.

Majestic's much larger index covering years of crawl data, including links that no longer exist. It is the right index for studying decay, for reconstructing what a site's profile used to look like, and for auditing a domain's past. My own link survival analysis is only possible because a historic index records links after they are gone.

Annotations that tell search engines which language and regional version of a page to show to which audience. In international link building it decides whether links earned for a local version support the right page. Hreflang is fiddly, must be reciprocal between versions, and is misconfigured on a striking proportion of the sites that use it.

A link from one page of a site to another page on the same site. Internal linking is the part of link work you control completely, and it decides how authority acquired from external links is distributed across your pages. Most sites have more to gain from fixing their internal linking than from acquiring another twenty external links.

Whether a page has been stored in a search engine's index and is eligible to appear in results. It is the first thing to check on any link prospect: a page that is not indexed cannot pass anything to you, and checking indexation takes seconds. A large proportion of purchased links sit on pages that were never indexed at all.

A service that distributes queries from reporters looking for sources, allowing experts to respond with quotes. Several such platforms operate, with varying quality of query. Response volume is high and the hit rate per response is low, so the tactic rewards discipline: answer only queries where you are genuinely the right expert, and answer them properly.

Google's structured database of entities and the relationships between them, assembled from many sources across the web. It underlies knowledge panels, entity disambiguation and a great deal of what search engines appear to know. Links and mentions both feed it, which is part of why a mention without a link is not worthless.

The map of the web as a network: pages and sites as points, links between them as connections. Search engines build and analyze this graph to work out which pages other pages consider important. Almost everything in link building is an attempt to change your position in it, and almost every link tool is a partial, independently crawled copy of it.

The complete set of links pointing at a site, considered as a whole: how many, from where, with what anchor text, on what kinds of pages, acquired over what period. Individual links matter far less than the shape of the profile. Audits, penalties and most diagnostic work operate at this level rather than link by link.

The cluster of sites that link to you, link to the same places you do, and link to each other. A site is judged partly by the company it keeps. Being surrounded by relevant, independently operated sites in your subject area is a good position; being surrounded by the same two hundred sites that link to every one of your competitors is not.

The informal name for the ranking value a link is thought to pass. Nobody outside a search engine can measure it, and treating it as a fixed quantity that gets divided up is a simplification of something considerably more complicated. It remains a useful shorthand when discussing why redirect chains, nofollow attributes and page pruning have downstream effects.

The rate at which a site gains new links over time. There is no correct velocity, and the idea that a particular pace triggers a penalty is folklore. What does matter is whether the pattern is explicable: a genuine news event produces a sharp spike, and so does a purchase, but the two look different in every other respect.

The steady loss of links over time as pages are edited, rewritten, migrated or pruned. It is the most underestimated force in link building. On my own site's Majestic data, the median referring domain stops linking after about 1,080 days and 49.6% of every link ever recorded has been lost, on a profile where nothing was ever bought.

A link a crawler previously recorded and can no longer find. The instinctive assumption is that the page died, but that is rarely what happened: on my own data, 97.4% of lost links disappeared while the source page was still perfectly reachable, and only four out of 645,202 were lost to a 404. Links are edited away, not deleted with their pages.

Something on your site that other people have a reason to cite: original data, a free tool, a definitive reference, a calculator, an unusually good explanation of a difficult subject. Every earned tactic depends on one existing. Most link building programs that fail were trying to acquire links to pages nobody would ever have a reason to reference.

Recovering links you already had and lost: pages that were redirected badly, URLs that changed in a migration, links that were removed during an edit, and citations that were never linked. It is the highest-conversion work in link building because the relationship and the editorial decision already existed, and given how much of any profile decays, there is always some.

Comparing your referring domains against those of several competitors to find sites that link to them and not to you. It produces a prospect list where relevance is already demonstrated and a willingness to link to companies like yours is already proven. The quality of the output depends entirely on choosing genuine search competitors rather than commercial rivals.

A listing of a business's name, address and phone number in a directory, map service, industry body or local publication, with or without a link. Citations support local search visibility through consistency rather than through link value, which is why the important question is whether the details match everywhere, not what the directory's authority score is.

An arrangement where two sites agree to link to each other, or where several sites rotate links among themselves to disguise the reciprocity. Excessive link exchange is named explicitly in Google's link spam policies. Two genuinely related businesses linking to each other because it helps their readers is not the problem; the systematic version is.

Google's term for arrangements intended to manipulate rankings through links, including buying or selling links that pass ranking value, excessive exchanges, large-scale article campaigns with keyword-rich anchors, and automated link creation. The definition is written broadly and deliberately, and it is the document worth reading before believing any vendor's claim that a tactic is compliant.

A model trained on very large quantities of text that generates language in response to a prompt. In search, models of this kind summarize retrieved documents and decide which to cite. They do not browse the web while answering unless connected to a retrieval system, which is why being present in an underlying index still matters.

An HTML instruction in the head of a page controlling whether it may be indexed and whether its links may be followed. Unlike robots.txt it operates per page and requires the page to be crawlable to be read at all. A noindex on a page that has earned links is worth checking before you conclude the links are not working.

A penalty applied by a human reviewer at Google rather than by an algorithm, reported in Search Console with a stated reason. Link-related manual actions can affect specific pages or a whole site, and lifting one requires cleaning up the links, documenting the effort and submitting a reconsideration request. Algorithmic suppression, by contrast, is never announced.

A value of the rel attribute, introduced to let publishers link to something without vouching for it. It was originally treated as a strict instruction not to pass ranking value; Google later described it as a hint, meaning the engine decides. Nofollow links still drive traffic, still build awareness, and still appear in the tools people use to judge you.

A link whose anchor text is the URL itself, displayed as plain text. Naked URLs appear constantly in real citation — in forum posts, references, comments and anywhere somebody pasted an address rather than writing a phrase — so a healthy profile contains plenty of them. They also carry no keyword information, which is precisely why they look natural.

A link added to an existing, already published article, usually by paying whoever controls the site. It is attractive because the page may already have age and links of its own. It is graded high risk here because the sentence containing your link was written after the fact, for money, and nobody made an editorial decision to cite you.

Attaching your commentary or data to a story already in the news cycle, quickly enough to be useful to journalists still writing about it. Speed is the entire tactic; a perfect response two days late is worth nothing. It requires a genuine subject-matter expert who can be reached and quoted within hours, which is why most organizations cannot execute it.

Whether a business's name, address and phone number appear in exactly the same form across every listing that mentions it. Inconsistent details make it harder for a search engine to be confident that all those listings describe the same real-world entity, and confidence is the thing local search visibility is built on.

The HTTP status returned when a requested URL does not exist. In link building a 404 on a page that has links pointing at it is wasted value, and finding those is the core of reclamation work. Notably, links are rarely lost to 404s: on my own data only four of 645,202 lost links went to one.

Deliberately pointing manipulative links at a competitor in the hope of causing them harm. Google has long stated that it ignores most such links rather than punishing the target, and in practice the panic it causes does more damage than the links do. Monitor your profile, keep records, and resist disavowing at the first sign of unfamiliar links.

The planned series of messages sent to one prospect: an initial approach and a small number of follow-ups, spaced deliberately. Most replies to cold outreach arrive after the first message rather than on it, so having no follow-up wastes the work of prospecting. Having six is harassment, and it damages the sending domain as well as the relationship.

A link acquired in exchange for money, goods, services or any other consideration. Paying for links that pass ranking value is named explicitly in Google's link spam policies. The compliant version is to disclose the arrangement with the sponsored attribute, which removes the ranking effect and leaves the traffic, the exposure and the audience.

The original algorithm for scoring pages by the quantity and quality of links pointing at them, where a link from an important page counts for more than a link from an obscure one. The public toolbar score was retired long ago, but link-based scoring did not go with it. Every modern link metric from every vendor is a descendant of the same idea.

Moz's equivalent score for an individual page rather than a whole domain. It matters because links are acquired on pages, not on domains: a strong site can have a page nobody links to and nobody reads. Checking the page-level score alongside the domain-level one prevents a great deal of wasted outreach.

Paying a wire service to distribute an announcement across a network of sites that republish it. The syndicated copies carry duplicate content and links that are, by definition, paid distribution rather than editorial coverage. It is useful for legally required disclosures and for reaching journalists. As a link tactic it is graded high risk here and produces little.

A group of sites controlled by one party for the purpose of linking to that party's clients or properties, usually built on expired domains that retain some historic authority. Networks are identified through footprints, and identification tends to be retrospective and collective: when a network is found, the sites relying on it are affected together.

The research phase of outreach: finding pages and people that represent a genuine opportunity, before any email is written. It is the half of outreach that determines the result, and the half that is skipped when a program is asked for volume. A qualified prospect list is the actual deliverable of good outreach work.

The proportion of outreach messages that result in a live link, which is always substantially lower than the response rate. It is the figure that determines how many prospects a campaign actually needs, and the one vendors are least willing to state. Measuring it honestly for your own program is what turns outreach from guesswork into a forecastable process.

An HTTP status code indicating a URL has moved for good. Search engines follow it and consolidate the old page's link signals into the new one, which is why it is the correct instrument for migrations and URL changes. A 302 declares a temporary move and does not consolidate the same way, and using one by mistake during a replatform is a common and expensive way to lose the value of links you spent years acquiring.

The Google algorithm update, first launched in 2012, that targeted link spam and manipulative anchor text at scale. It changed the economics of link building permanently by making bad links a liability rather than merely a waste of money. Link spam handling was eventually folded into the core ranking systems, so the name now describes an era as much as a system.

A unique website that links to you at least once, regardless of how many individual links it sends. Referring domains are almost always the more meaningful count. On my own site's data, one domain accounted for 219,159 links through a single sitewide link, which is 16.8% of the entire link total but exactly one referring domain.

The IP address of the server hosting a linking site. Counting distinct referring IPs alongside distinct referring domains gives a rough sense of how independent your linking sites really are. Shared hosting means unrelated sites often share an address, so a low IP diversity is a prompt to investigate rather than proof of anything.

A page whose purpose is to list useful links on a subject, maintained by a library, an association, a university department, a hobbyist or a company. Resource pages are one of the few places on the web where somebody is actively looking for things to link to, which makes them a durable outreach target and an unusually receptive audience.

The proportion of outreach messages that receive any reply, positive or otherwise. Published figures vary by an order of magnitude depending on who is publishing them and what they are selling, so the only number worth trusting is your own, measured consistently. Track it by segment; a low overall rate usually hides one segment performing badly.

A sequence where one URL redirects to another, which redirects to another, before reaching the final page. Chains accumulate over years of migrations. They slow crawling, waste crawl activity, and add opportunities for a step to break — which severs every link pointing at the front of the chain. Auditing them is standard work when inheriting a site.

A plain text file at the root of a domain that tells crawlers which paths they may request. It controls crawling, not indexing, and the distinction matters: a page blocked in robots.txt can still appear in results, and blocking a page prevents a crawler from seeing the links on it or the redirect it issues.

The technique of retrieving documents from a search index or database and giving them to a language model so that its answer is grounded in specific sources. It matters for link building because it explains the mechanism: to be cited in an answer, your page must first be retrievable, and then must be the clearest available statement of the point.

Visitors who arrive by clicking a link on another site. It is the only effect of a link you can measure directly and completely, without relying on any third-party estimate, and it is routinely ignored in favor of metrics that are guesses. A link that sends real visitors is worth having whatever attribute it carries.

The rel value used to mark links that were paid for or that form part of an advertising or sponsorship arrangement. It is the disclosure mechanism Google asks publishers to use for commercial links, and applying it correctly is what separates a legitimate paid placement from an undisclosed one. Applying it also removes any ranking benefit, which is the point.

A link that appears on every page of a site because it sits in a template: a footer, a sidebar, a navigation bar, a theme credit. One sitewide link can generate an enormous raw link count from a single referring domain. Search engines have long compressed the value of repeated template links, so the count flatters a profile far more than the signal does.

Finding content that has attracted many links, producing something substantially better on the same subject, and contacting everyone who linked to the original. It works when the new resource is genuinely superior and the outreach is targeted. It fails, constantly, when better means longer, and the resulting emails are indistinguishable from ordinary outreach spam.

A link acquired by offering a scholarship in order to be listed on university financial-aid pages. The pattern is well known, the pages are frequently pruned once the practice is recognized, and the scholarship exists as a pretext for the link. It is graded high risk on this site despite the .edu domain that makes it look respectable.

Submitting URLs to public bookmarking and aggregation services to generate links. The links such services produce are nofollowed, user-generated and produced at volume by anybody who wants them, which is exactly why they stopped counting for anything. Graded obsolete here, and still bundled into cheap packages because it costs nothing to deliver.

Machine-readable markup added to a page that states explicitly what the page and its contents are: an article, a person, a product, an organization, a FAQ. It does not directly improve rankings, but it removes ambiguity about your identity and your claims, which matters increasingly as systems select sources to summarize and cite.

The shared vocabulary used for structured data, maintained collaboratively by the major search engines. It defines the types and properties — Organization, Person, Article, FAQPage and so on — that markup uses. Using the vocabulary correctly is more valuable than using a great deal of it, and inaccurate markup is worse than none.

A structured data property that points at other authoritative URLs describing the same entity: an official profile, an encyclopedia entry, an industry registry. It is the plainest available way of telling a machine that several identities across the web are one thing, and it is underused compared with how much disambiguation work it does.

The proportion of relevant conversations, results or generated answers in which your brand appears, relative to competitors. It is a more honest measure of a link and PR program's effect than a raw link count, because it captures visibility that arrives without a hyperlink attached — which is an increasing share of the total.

The search engine results page: everything returned for a query, including organic results, advertisements, map packs, featured snippets and generated answers. Reading the results page for your target queries tells you what kind of page can rank, which is a more useful input to tactic selection than any keyword difficulty score.

A link that points directly at your own site, as opposed to one pointing at a page that links to you. The term only exists because of tiered schemes, but the concept is useful in analysis too: describing a link as tier one specifies its distance from you in the graph without implying anything about how it was acquired.

Examining not just the sites that link to you but the sites that link to those sites, out to several removes. It reveals where authority in your market actually originates, which of your referring domains are themselves well supported, and whether a cluster of links is genuinely independent. It is an analytical skill, and it has nothing to do with tiered link building.

A scheme in which low-quality links are built to the pages that link to you, on the theory that strengthening those pages strengthens you indirectly. It is a link scheme with an extra step, produces detectable patterns at every level, and is graded high risk on this site. It is routinely confused with tiered link analysis, which is a completely different activity.

Majestic's measure, on a 0 to 100 scale, of how trustworthy the sites linking to a page or domain are. It is calculated by distance from a manually reviewed seed set of trusted sites, so it rewards proximity to well-established sources rather than raw link volume. It is the metric I use first when judging whether a referring domain is worth anything.

Majestic's breakdown of Trust Flow by subject category, showing not just how trusted the linking sites are but what they are trusted about. A site with high Trust Flow in a category unrelated to yours is a much weaker prospect than the headline number suggests, which makes this the single most useful field for judging relevance quickly.

Trust Flow divided by Citation Flow. A high number means the links pointing at a site come from comparatively trustworthy places; a low number means there is a great deal of link volume from sources that are not trusted. It is a fast first filter when reviewing prospects, because manufactured profiles almost always have plenty of volume and very little trust.

A vendor term for a link a tool considers harmful, produced by an automated score rather than by any statement from a search engine. The label is far more confident than the underlying analysis justifies, and the tools that produce it frequently also sell the cleanup. Judge links yourself before acting on a score generated by something that profits from alarm.

The extent to which a site is treated as a substantive source on a specific subject, rather than as generally strong. It is built by covering a subject thoroughly and by being cited by others working in that subject, which is why relevance beats raw authority scores when choosing which links to pursue.

The rel value for links inside user-generated content: forum posts, blog comments, profile fields, community submissions. It tells a search engine that the site owner did not write or vet the link. Most large platforms apply it automatically across every link users can create, which is one reason tactics built on user-generated links stopped producing ranking value.

Ahrefs' page-level equivalent of Domain Rating, scoring the link strength of one specific URL. As with every page-level metric, it is the more relevant figure when you are evaluating whether a particular page is worth pursuing a link on, since domain-level scores tell you about the site's best pages rather than the one in front of you.

A reference to your brand, product, research or people in somebody's published content, with no hyperlink attached. Historically these were treated as leads for reclamation outreach. They increasingly matter in their own right, because systems that summarize and cite sources work from named references rather than only from hyperlinks.

A free hosted page on a blogging or publishing platform, created solely to carry links back to a target site. The pages have no audience, no editorial oversight and no purpose beyond the link, and platforms apply nofollow or UGC attributes across user-created links as a matter of course. It is documented here so readers recognize it in a proposal.

Shorthand for tactics that comply with published search engine guidelines. The label is less useful than it sounds, because it is a moral vocabulary applied to an engineering question and because almost everyone claims it. This site grades tactics as safe, situational, high risk or obsolete instead, which forces a specific judgment rather than a badge.

The Internet Archive's public record of archived web pages over time. In link building it is used to see what a dead page contained before it disappeared, to reconstruct a domain's past before purchase, and to prove what a page said on a given date — which is also why it appears constantly in investigations and expert witness work.

A machine-readable file listing the URLs a site considers worth crawling, optionally with modification dates. It helps discovery, particularly on large sites and for pages with few internal links, but it is a suggestion rather than an instruction. A sitemap will not make a page rank and will not compensate for a page nothing links to.