Backlink Health 101: DoFollow Ratio & Referring IPs | Adviora

Backlink Health 101: DoFollow Ratio, Editorial Links and Referring IPs

Every backlink tool hands you a large number within a minute of connecting your domain. Far fewer tell you what it means, or which half of it to ignore. A useful backlink profile analysis is not a bigger total - it is a set of precisely defined metrics, each answering one narrow question and refusing to answer the others.

Link metrics are the most misread numbers in SEO. Practitioners chase a DoFollow ratio as though it were a target. Founders quote a Domain Rank as though Google published it. Teams disavow links that were never hurting them. Each mistake has one root cause: treating a descriptive statistic as a lever.

This guide defines each core link metric, states what it does and does not tell you, and separates vendor inventions from rules Google actually published. Adviora's Backlink Report shows them in one view

Why Backlink Health Matters in 2026

Links are one of the few signals a site cannot generate on its own. That independence is what makes them useful and what makes them a target. Google has never published a public link score, and says plainly that no special optimisations are required to appear in AI Overviews or AI Mode beyond being indexed and snippet-eligible. The 2026 job is not a new tactic - it is to stop misreading what you already have.

• Cheap link acquisition makes raw totals less informative than they used to be.
• Migrations and replatforms silently break inbound links, so a profile decays untouched.
• Vendor spam scores tempt teams into disavowing links that took years to earn.

Adviora's Backlink Report puts those realities on one screen instead of four tools.

Total Backlinks vs Referring Domains: The Comparison That Matters

Total backlinks counts every individual link pointing at your domain. Referring domains counts the unique websites those links come from, and the gap between the two is the most informative thing on the page. A site-wide footer link produces one link per page on the partner's site: thousands of backlinks, one referring domain.

• Total backlinks - volume, inflated by footers, blogrolls, templates and paginated archives.
• Referring domains - breadth: distinct sites that decided, at least once, to point at you.
• New and lost referring domains per month - the only link trend worth reporting upward.

If a number can be multiplied by someone else adding pages to their own site, it is a volume metric. Read referring domains first, and use total backlinks only to explain the gap.

DoFollow vs NoFollow: A Description, Not a Target

A DoFollow link is simply one with no rel attribute restricting it. The alternatives are annotations the linking site adds to describe the relationship.

• rel="nofollow" - the linking site declines to endorse; Google names this acceptable marking for paid links.
• rel="sponsored" - for paid or compensated placements, also named acceptable by Google.
• rel="ugc" - user-generated content such as blog comments and forum posts.

Profiles built from real coverage pick up nofollowed mentions from press sites, comment threads and syndication along the way. That describes how coverage happens; it is not a threshold. The failure mode is engineering the ratio - buying nofollowed links to balance a profile, or refusing a valuable placement because it moves the number. Read the ratio as a description of your coverage, nothing more.

Editorial Links vs Placed Links, and What Google Calls Link Spam

An editorial link is one somebody chose to give you because your page helped their reader. A placed link is one you obtained - by payment, exchange, guest post or automated tool. The distinction is not in the HTML, it is in who decided. Google's spam policies use that framing: link spam is creating links to or from a site primarily for the purpose of manipulating search rankings.

• Named by Google as link spam: buying or selling links that pass ranking signals, excessive link exchanges, automated link creation, and undisclosed paid content.
• Explicitly acceptable: paid links marked rel="nofollow" or rel="sponsored". The problem is never payment - it is undisclosed payment that passes ranking signals.
• Not on Google's list: a low DoFollow ratio, links from one country, or a modest Domain Rank.

Adviora's Backlink Report counts editorial links separately, so you can see how much of the profile you earned rather than assembled.

Referring IPs and Subnets: A Fraud Signal, Not a Quality Signal

A referring IP is the server address hosting a linking site; the subnet is the block it sits in, usually compared at class C level. Most guides read this backwards. IP diversity is a fraud signal, not a quality signal: low diversity is evidence something might be wrong, high diversity is not evidence anything is right.

• Many referring domains on one IP - could be a private blog network, could be cheap shared hosting. Open them before concluding.
• Many domains in one class C subnet, linking to you, registered around the same time - a much stronger network indicator.
• Wide IP spread across a profile of thin directories - still a thin profile. Diversity fixed nothing.

Listing referring IPs beside the domains turns clustering into one sort rather than a spreadsheet exercise.

Anchor Text and Spam Score: Two Patterns, Neither a Verdict

Anchor text is the visible clickable text of an inbound link. No search engine publishes a safe percentage for any anchor category, and anyone quoting one is quoting a vendor model. Spam score is likewise a vendor-calculated likelihood - a model output, not a penalty notice.

• Branded and naked-URL anchors normally dominate a profile built from real coverage.
• Generic and partial-match anchors are unavoidable in editorial writing and not worth analysing.
• Exact-match commercial anchors repeated across unrelated domains in a short window reliably read as bought.
• Worth disavowing: links you or an agency bought, hacked-site injections, traceable network placements.

Disavowing is a one-way door, so the bar should be evidence rather than a colour on a dashboard. Adviora produces disavow candidates as a review queue, not a one-click export.

Domain Rank and Domain Authority Are Vendor Composites, Not Google Metrics

This deserves stating without hedging. Google publishes no public authority score for websites. Every 0–100 site-strength number - Domain Rank, Domain Authority, Domain Rating, Authority Score, including the Domain Rank in Adviora's own Backlink Report - is a vendor composite modelled from that vendor's crawl of the link graph.

• Not comparable across vendors: the same site scores differently because crawls and models differ.
• Relative, not absolute: useful for rank-ordering prospects, near-meaningless as a KPI.
• Gameable, which is precisely why cheap link sellers advertise them.

Use them as an analyst uses a screening score: sort the list quickly, then open the file and read it. Any dashboard that reports one should label it a composite rather than dress it up as a Google signal.

Best Practices for Reading a Backlink Profile

1. Open referring domains first; treat total backlinks as context, never the headline.
2. Chart new and lost referring domains monthly - direction beats absolute size.
3. Check the editorial link share before celebrating any growth number.
4. Sort by referring IP quarterly and open anything that clusters.
5. Treat spam score as a review queue, and log the evidence behind every disavow decision.
6. Fix broken inbound links before chasing new ones - recovered links are cheaper.

Run that list monthly and link work becomes maintenance with an audit trail.

The Future of Link Evaluation

The direction of travel is away from counting and towards corroboration. As answer engines assemble responses from several sources at once, the useful question shifts from how many sites link to you towards whether the sites describing your category treat you as a reference point. Nobody publishes a metric for that yet.