What is link building?

By Bill Hartzer · Hartzer Consulting · 25+ years in search, domains and expert witness work

What link building actually is

Link building is the practice of getting other websites to link to yours. That is the entire definition. Everything else — the tactics, the tools, the arguments about what is legitimate and what is spam — hangs off that one sentence.

A link is a hyperlink: a piece of text or an image on someone else's page that, when clicked, sends a reader to a page on your site. Search engines follow those links to discover pages they have not seen, and they use the pattern of links across the whole web as one input into deciding which pages to rank. A link from another site is called a backlink, and the site it came from is a referring domain. Those two words are not interchangeable and the difference matters more than almost anything else in the field: one site can send you a hundred thousand links, and it still counts as one opinion.

I want to be blunt about the phrase itself. Link building is a slightly dishonest name for the work, because the links you cannot build are the ones worth most. Nobody constructs a citation from a national newspaper. You do something worth citing, and somebody cites it. The parts you genuinely build are the smaller, more mechanical links — a directory entry, a resource page listing, a fixed dead link — and those are useful but they are not what moves a competitive result.

Why search engines count links at all

The idea is older than most people working in SEO today. In 1998 the PageRank paper described a way of ranking web pages by treating a link as a vote, where votes from pages that themselves receive many votes are worth more. It was a solution to a specific problem: in the mid-1990s, search engines ranked pages largely on the words in the page itself, which meant anybody could write their target phrase two hundred times in white text and win. Links were the first widely used ranking signal that came from outside the page being ranked, and therefore the first that the page's owner could not simply declare.

That is still the reason links matter. A link is an independent party spending a small amount of their own credibility on you. It is evidence of association that you did not author. Search engines have added an enormous amount since — content quality assessment, user behavior signals, entity understanding, freshness, and much that is not public — but the structural appeal of a citation you did not write yourself has never gone away.

It also explains why link spam exists and why Google spends so much effort on it. The moment a signal is valuable and comes from outside, someone will try to manufacture the outside. Penguin, launched in 2012, was the first large public attempt to devalue manipulated links at scale rather than merely penalize individual sites. It changed the economics of the field permanently.

The main families of tactic

Every link building tactic I have ever seen falls into one of three families, and knowing which family you are in tells you most of what you need to know about the risk, the cost and the ceiling.

• Earned links. Somebody chose to cite you without being asked. Digital PR, original research, linkable tools and assets, expert commentary, podcast appearances, conference coverage. Highest ceiling, highest skill requirement, least controllable.
• Outreach links. You asked, and a human editor said yes. Broken link building, resource pages, local and community links, unlinked mention reclamation, competitor gap work, guest posting. Moderate ceiling, high labor cost, entirely defensible when the ask is honest.
• Placed links. You or a vendor put the link there, usually with money or an automated process. Paid links, niche edits, private blog networks, tiered schemes, syndicated press releases, forum profiles. Fast, scalable, and the family Google's link spam policies exist to target.

The reference section of this site grades every tactic on four risk levels — safe, situational, high-risk and obsolete — which I find far more useful than the white hat and black hat vocabulary. Safe means the mechanism does not violate anything even when done badly. Situational means it is fine at honest volume and becomes a footprint at scale. High-risk means it is against the published policies and the only question is detection. Obsolete means it once worked and now does nothing, which is its own kind of waste.

What link building can actually do

Links do three things reliably, and it is worth separating them because people conflate all three into a vague hope about rankings.

Discovery. A crawler that finds a link to a page it has never seen will usually go and fetch that page. For a new site with no history, links are still the most reliable route into an index. This effect is real and fast and nobody argues about it.

Ranking. Links contribute to how pages are ordered, and the size of the contribution varies enormously by query. On a competitive commercial term where a dozen sites all have adequate content, link difference is often what separates them. On a long-tail informational query with thin competition, a good page with almost no links routinely wins. Anybody who tells you links are worth a fixed percentage of the ranking decision is selling something.

Referral and association. A link on a page people actually read sends people. It also puts your name next to a subject in a place a machine can read, which increasingly matters for how systems that summarize the web decide who to name. That last part has grown in importance in the past few years and is the subject of the final module of the course.

What link building cannot do

This is the part most introductions skip, and it is where the money gets wasted.

Links cannot rescue a page that does not answer the query. If the intent behind the search is transactional and your page is a blog post, no volume of links fixes the mismatch. I have watched six-figure campaigns fail on exactly this.

Links do not stay. This is the single least understood fact in the discipline. On a twenty-year profile I analyzed in detail — 1,301,839 links from 22,260 referring domains, none of them ever bought — the median referring domain stopped linking after 1,080 days, and 49.6% of all links ever recorded had been lost. Even the good ones erode. A link from a Trust Flow 61+ domain survived a median of 3,353 days, roughly four times as long as a link from a Trust Flow 0 domain at 859 days, but four times longer is not forever. Link building that stops is a program losing ground.

Links do not arrive on a schedule. On that same profile, only 446 referring domains — 2.0% of the total — reached Trust Flow 41 or above. Over twenty years that is about twenty-two genuinely strong domains a year, from a site publishing constantly. Any proposal promising you thirty authoritative links a month is describing something that does not exist in nature.

Links cannot make you interesting. The hardest input is having something worth pointing at. No tactic substitutes for it.

Where to start, honestly

If you have arrived here cold, the useful sequence is short. First, understand what a link is doing and stop counting raw backlink totals — count referring domains instead, because the totals are dominated by sitewide links. On the profile above, one single domain accounted for 219,159 links, 16.8% of everything, from one sitewide placement. The top hundred domains accounted for 84.3% of all links. Raw counts describe almost nothing.

Second, learn to judge a link before you chase it. Relevance first, metrics second, and always check whether the linking page is indexed and alive.

Third, pick one tactic that fits your site rather than five that fit somebody's case study. A local service business and a software company with a research budget should be doing completely different work.

The nine-module course on this site teaches that sequence properly, starting with how links work and ending with what changes as search shifts toward summarized answers. The tactic reference covers each individual method in depth, with a risk grade and a blunt verdict on each. This page is the map; those are the territory.