Not All Links Are Equal: the Different Types of Backlinks in SEO

The industry talks about “quality links” as if there’s one definition. There isn’t.

A brand mention on Reuters and a guest post on a niche blog aren’t competing for the same job. They target different pages, pass different signals, and produce different outcomes.

Treating them as interchangeable — or building only one type — is one of the most common ways link building goes wrong.

This post maps the full spectrum: what each link type does, where each one fits, and what the leaked ranking data from Yandex and Google shows about how different links are actually weighted.

The Four Types and What They Actually Do

Link types aren’t a hierarchy of quality. They’re a map of function. Each one does something specific, and the strategy question isn’t “which is best?” Rather, it’s “which part of the spectrum am I missing?”

Let’s answer that now:

Brand and HARO Links

Target: homepage and brand pages.

These build domain trust and entity recognition. Coverage on major publications tells Google and AI systems that your brand is a real, recognized entity in its space. That signal operates at the domain level, aka it lifts the site as a whole rather than moving individual pages.

The trade-off: you can’t point these at deep service pages. That’s not their job. Their job is to establish the foundation on which everything else sits. They can be slow and sometimes expensive to earn, but they’re also the hardest for competitors to replicate quickly, so they offer a decent moat.

Guest Posts

Target: service pages and product pages.

Guest posts are where page-level ranking gets built. You control the target URL and the anchor text, which makes them the right tool for moving specific pages on competitive terms. A well-placed guest post on a relevant, high-traffic site with a precise anchor pointing to the right service page is direct page-level ranking work.

Slight caveat: The production cost is higher. You need the content, the outreach, and the placement. But the control is worth it for revenue-driving pages.

Niche Edits

Target: informational pages where contextual fit is natural.

Niche edits skip the content production step. The page already exists, already has authority, and already ranks. You’re adding a link to it. That makes them faster than guest posts and cheaper to execute.

The trade-off is less control. You don’t own the surrounding content, so relevance depends on how well the placement is matched.

Volume Links

Target: anywhere, but these are a supplement, not a foundation.

Volume links add diversity to a link profile. They’re the support layer. They fill gaps and round out the profile once the higher-tier work is in place. The mistake is treating volume as a strategy rather than a component.

The SEO graveyard is full of sites with hundreds of volume links and no movement on competitive terms, because the domain trust layer was never built.

Which Link Type Goes to Which Page

Most link-building guides skip this part. The type of link determines where it should go, and getting it wrong can water down its impact on authority and rankings.

We’ve touched on it in the previous section, but to reiterate:

Brand and HARO links go to the homepage and brand pages. They can’t be pointed at deep service pages. Their job is domain-level authority, and that’s where the signal lands.

Guest posts go to service and product pages. You have full control over the target URL and anchor, which is exactly what page-level ranking requires.

Niche edits go to informational pages. The contextual fit has to be natural, and informational content provides the right environment for that.

Two mistakes come up regularly: either all guest posts point to the homepage with brand anchors, or every link points to service pages with exact-match anchors.

Both are incomplete. The first wastes the control guest posts give you. The second skips the domain trust work that makes everything else more effective.

What Yandex and Google Leaks Show About How Different Links Are Weighted

Until very recently, search engines, particularly Google, which is known for keeping its cards close to its chest, operated as a black box. Meaning, we were all optimizing around data collected over time or, much worse, pure, unadulterated speculation.

But two leaks have given the industry its clearest view yet of how search engines actually score links: the 2023 Yandex source code leak and the 2024 Google API leak.

The findings change how you should think about link selection.

From the Yandex Source Code

When the Yandex source code leaked, we got our first true glimpse behind the search engine curtain, giving us all a hint at how Google’s search algorithms might be behaving, too.

The entire classic link analysis module in Yandex — 60 factors — was deprecated. All of it. The system replaced it with BrowserPageRank: link signals validated by actual user behavior.

A link from a page nobody visits produces no behavioral data, which means it carries no weight in the modern system. Traffic to the linking page is how the infrastructure generates the signals it actually runs on.

The coefficients are specific. Yandex’s link scheme penalty factor (FI_COMM_LINKS_SEO_HOSTS) sits at –0.181. Google’s PageRank’s positive contribution is +0.183.

The numbers are almost identical in magnitude but run in opposite directions. The system is designed to cancel out the gains from link manipulation at the same weight as those gains would help.

On placement: content-embedded links carry a coefficient of +0.094 (FI_NUM_LINKS_FROM_SEGMENT_CONTENT). That’s roughly half the weight of PageRank.

Editorial placement in a paragraph is measurably more valuable than a blogroll, sidebar, or footer link. Where your link lives on the page matters in the data.

From the Google API Leak and the DOJ Trial

The 2024 Google API leak exposed the actual attributes Google uses to evaluate links, named variables pulled directly from the ranking codebase. They’re internal signals that third-party tools don’t have access to.

The attributes break down into four categories:

Trust signals:

• siteAuthority: Google’s own domain-level trust score, separate from PageRank. The higher a site’s siteAuthority, the more ranking power a link from it carries.
• homepagePageRankNs: The PageRank of the linking site’s homepage. Even when a link comes from a deep inner page, Google factors homepage authority into its evaluation.

Relevance signals:

• siteFocusScore: How topically coherent the linking site is. A focused niche site scores differently from one that covers everything and nothing.
• topicEmbeddingsVersionedData: Semantic relevance scored between the linking site and your target page. A link from a site operating in the same topic space carries more weight than one from an unrelated domain.

Placement and dilution signals:

• titlePageRank: PageRank calculated specifically for the main content area of a page. A link placed editorially in the body of a relevant article carries more weight than one in a footer, sidebar, or navigation.
• pagerankWeight: The actual per-link weight assigned in Google’s PageRank calculation. A link on a page with dozens of outbound links passes less than one on a page with fewer.

Penalty signals:

• anchorMismatchDemotion: A confirmed demotion signal. If the anchor text doesn’t match the content of the target page, the link can actively work against you.
• scamness: A scam detection score ranging from 0 to 1023, applied directly in ranking. A link from a high-scamness site carries those signals with it. DR has no equivalent.

The attributes confirm what many in the industry had theorized about for years, but couldn’t prove. The DOJ antitrust trial went a step further: it put Google’s own people under oath and produced the internal documents to back it up.

During the trial, Google’s VP of Search testified under oath that NavBoost — their behavioral click signal system — is more powerful than any of their deep learning models for ranking.

The internal documents state it directly: “Learning from logs is the main mechanism behind ranking.” A link from a page users actually visit produces behavioral data. A link from a page nobody visits doesn’t. That gap is the difference between a link that works and one that doesn’t.

Links Now Feed Two Systems, Not One

Google confirmed in internal documents that they use search authority signals to filter what enters Gemini’s pre-training data. Of the 160 billion tokens evaluated, 80 billion were removed based on quality signals, including link-derived authority.

The same links building your Google rankings are determining what Gemini — and every Gemini-powered product — learns about your brand. It’s not two separate investments. It’s one investment serving two systems.

The data supports this: brand mentions correlate with AI citation rates at 0.664 (Spearman’s rank correlation). Referring domains correlate at 0.218. The gap is significant. Domain trust and entity recognition — the job that brand and HARO links do — translates more directly to AI visibility than raw link count.

This is why the top of the spectrum matters more now than it did two years ago. Brand links build the foundation for Google rankings and feed the entity signals that AI systems use to understand who you are.

Building the Architecture: Top-Down, Not Bottom-up

The leaked data reinforces a principle that the best link builders have operated on for years: the order in which you build matters as much as the types you build.

Start with the foundation. Brand links establish domain trust and entity signals. Without this layer, everything above it is harder. These are slow, expensive, and can’t be scaled on demand, but they raise the ceiling on what everything else can achieve.

Build the core. Once the domain has trust, guest posts, and niche edits, move specific pages on specific terms. This is where the page-level ranking work gets done. The anchor precision and target URL control that guest posts give you only work well when the domain is already trusted.

Add the support layer. Volume links fill out the profile and add diversity. They’re the last step, not the first. The ratio changes by site and vertical. The order doesn’t.

The most common failure pattern: clients who’ve built hundreds of volume links and can’t compete for competitive terms. The foundation wasn’t there. The spectrum explains why.

The Full Spectrum, Covered

The strategy is rarely missing links. It’s missing the right type of links pointing to the right pages.

Loganix’s product line covers each part of the spectrum:

1. Brand Links: domain trust and entity signals
2. HARO-Style Link Building: media placements from major publications
3. Guest Posts: page-level ranking with full URL and anchor control
4. Niche Edits: speed and contextual relevance on existing pages
5. Shop the List: self-serve across 34,627 pre-vetted sites, filtered by traffic, topical relevance, and quality signals
6. AI Brand Links: brand mentions and entity signals built for AI citation architecture

Not sure where to start? Enter your URL in Shop the List, and we’ll match you with the sites most relevant to your domain.