Backlinks: what they are, how to earn them, what to avoid

A backlink is a link from another website to a page on yours. Search engines read it as a vote. This guide is for founders and marketers who want more of those votes without risking a Google penalty.

What is a backlink, and where did the idea come from?

A backlink is any link on another site that points to yours, and its value depends on who is linking and why. The idea comes straight from academic citations.

In 1998, Sergey Brin and Lawrence Page published The Anatomy of a Large-Scale Hypertextual Web Search Engine in the journal Computer Networks and ISDN Systems. They described PageRank as "an objective measure of its citation importance". Unlike simple citation counting, PageRank did not count all links equally: a link from an important page was worth more, and a page's vote was split across all the links on it.

The paper explained it with a "random surfer" who keeps clicking links. The chance that the surfer lands on a page is its PageRank. The paper also said Google treated link text, the anchor text, as a description of the page being linked to, because "anchors often provide more accurate descriptions of Web pages than the pages themselves".

Why do backlinks still matter now?

Because Google still uses them, though less crudely than in 1998. Google's ranking systems guide (last updated 10 December 2025) says it has "various systems that understand how pages link to each other", and calls PageRank one of its core ranking systems, one that has evolved since Google launched.

What changed is how Google handles links it does not trust. On 14 December 2022, Google said its AI-based system, SpamBrain, would "neutralize the impact of unnatural links", and could now detect "both sites buying links, and sites used for the purpose of passing outgoing links". Many bad links are now simply ignored, so money spent on them is wasted.

How does Google read nofollow, sponsored and ugc?

Google asks site owners to label links they do not vouch for. Its link best practices say paid links should carry rel="sponsored" or rel="nofollow", links in user-generated content such as forums should carry rel="ugc" or rel="nofollow", and nofollow should be used "only when you don't trust the source".

The big change came in September 2019. In Evolving "nofollow", Google introduced sponsored and ugc and said all three attributes are now "treated as hints about which links to consider or exclude within Search". For crawling and indexing, nofollow became a hint from 1 March 2020. A nofollow link is no longer a hard block.

A link you had to pay for is an advert. A link someone chose to give you is a reputation.

How do you earn backlinks without breaking the rules?

You make something other people want to cite, then tell the people who write about your subject. Four methods work well, in this order of effort.

1. Publish original data

Journalists need numbers they can quote. A customer survey, anonymised enquiry data or a yearly price index gives them a reason to link to you as the source. Publish the method, sample size and date. This also suits AI answers: the GEO study by Aggarwal et al. found that adding statistics and citing sources improved visibility in generative answers by roughly 30 to 40% on its main measure.

2. Run digital PR around a real story

Digital PR means pitching that data, or an expert view on the news, to journalists and creators so they cover it and link to you. Keep pitches short, relevant to the reporter's beat and free of sales language.

3. Build a free tool or resource

A calculator, template or glossary attracts links for years. An EMI calculator for a lender or a size guide for a fashion brand gets linked to without anyone asking.

4. Write genuine guest contributions

Guest articles are fine when they inform another site's readers. Google's 2017 reminder on large-scale article campaigns said the problem is articles written mainly to build links back to the author's site, especially with keyword-rich anchor text across many sites. Write for sites your customers read, and let the editor decide the links.

When should you do each step?

A plan I would run for a mid-sized brand:

• Month one: audit existing links in Google Search Console, fix broken pages that already have links, and pick one dataset you can publish.
• Months two and three: publish the data with a clear method page, then pitch it to ten relevant journalists and trade sites.
• From month four: add one useful tool or resource each quarter, and one or two guest pieces for publications your buyers read.

What link tactics should you avoid?

Avoid anything where the link exists mainly to move rankings. Google's spam policies (last updated 28 August 2026) define link spam as "creating links to or from a site primarily for the purpose of manipulating search rankings". The examples include:

• Paid links that pass ranking credit, including money, goods or services for links, or "sending someone a product in exchange for them writing about it and including a link". Marked with sponsored or nofollow, the same deal is fine.
• Excessive link exchanges ("link to me and I'll link to you") and partner pages that exist only to cross-link.
• Automated links, low-quality directories and bookmark sites, keyword-rich links in widgets, and links spread across many sites' footers.
• Advertorials and guest posts with optimised anchor text that pass ranking credit.
• Private blog networks (PBNs). Google's policy does not use the term, but a network of sites built to pass links is what it means by "low-value content primarily for the purposes of manipulating linking and ranking signals", and what SpamBrain was trained to spot.

What is site reputation abuse?

It is renting out a trusted site's ranking power to third-party content. Google's policy targets third-party pages published on a host site "mainly because of that host's already-established ranking signals", such as a medical site hosting a casino page. On 19 November 2024, Google clarified that "no amount of first-party involvement alters the fundamental third-party nature of the content". Buying a "sponsored article" on a big news site's coupon or review section, hoping its authority rubs off, now carries real risk for the host and wastes the buyer's money.

Do backlinks matter for AI answers and GEO?

Yes, but brand mentions seem to matter more than raw link counts. AI answer engines draw on pages from search indexes, so the authority links help build still feeds them. Google's AI features documentation says a page must simply be indexed and eligible for a snippet to appear in AI Overviews.

The best evidence so far is correlational. Ahrefs studied 75,000 brands and published in May 2025 that branded web mentions had a correlation of 0.664 with AI Overview brand visibility, against 0.295 for referring domains and 0.218 for backlinks. Ahrefs warns that correlation is not causation. A second Ahrefs study, updated in March 2026, found only 37.9% of AI Overview citations came from pages in the top ten for the same query, down from about 76% in July 2025. AI answers now cite pages that rank well for related "fan-out" questions too.

My reading: links still help you get indexed and ranked, while being named and cited across the web matters more for being picked by AI. Digital PR does both. For the wider picture, see my guide to SEO, GEO, AEO and engine optimisation.

Where does link building go wrong in India?

It goes wrong when brands buy "guest post packages" by the hundred. Paid posts with do-follow links on general news and blog sites are exactly the paid-link pattern Google describes, and are likely to be ignored.

There is a disclosure rule too. The Advertising Standards Council of India's draft influencer guidelines of February 2021 said a paid blog post should carry its disclosure label in the title. So a paid placement needs a label for readers and a sponsored tag for Google.

The opportunity is regional and trade media: local news sites, trade associations and city guides that national competitors ignore.

Who should do this work?

The person who owns PR or content should own link earning, with the SEO lead checking quality. In practice:

• PR or communications pitches journalists and builds media relationships.
• Content and data teams produce the research, tools and guest articles.
• The SEO lead monitors links in Search Console, flags spam and keeps anchor text natural.
• Founders and subject experts give quotes and interviews.

Be wary of anyone who sells links by volume or guarantees a number of do-follow links a month.

My view

My view is that backlinks are a by-product of reputation, not a thing to buy. In my agency years at Ogilvy, Law & Kenneth and Percept Hakuhodo, and later running PR and reputation work in-house, the coverage that lasted came from having something worth saying. The same holds online.

The trade-off is speed. Earned links are slow and hard to forecast; a paid package shows fifty links next week. I would still take ten links from sites my customers read over five hundred from sites nobody visits, because the ten also bring traffic, trust and AI mentions, and they will not vanish in the next spam update.

Start with one dataset

Look at the data your business already has: enquiries, prices, customer questions, seasonal demand. Pick one finding that would surprise people in your industry, publish it with a clear method, and send it to five journalists who cover your sector. That single page is the start of a link profile you never have to hide.