A backlink audit checklist is a structured set of checks you run against your link export to determine whether the profile is healthy, risky, or carrying links that no longer exist. These 12 checks cover the signals that matter: whether your links are live, correctly attributed, topically relevant, naturally acquired, and free of patterns that map to Google’s spam policies. The goal is not to find reasons to disavow everything. It is to understand what you actually have, so you can protect the profile you have already built. For a full walkthrough of the audit process from setup to remediation, see the full backlink audit walkthrough.
The 12 Checks at a Glance
| # | Check | What Good Looks Like | Red Flag |
|---|---|---|---|
| 1 | Referring domain count vs. raw link count | Diverse sources; no one domain accounts for 20%+ of links | 80%+ of links from a handful of domains |
| 2 | Live vs. lost link status | Most links return 200; known losses are expected | Large share of 404s, redirects to unrelated pages |
| 3 | Attribute mix (dofollow/nofollow/sponsored/ugc) | Mix reflects natural acquisition; sponsored links properly tagged | 95%+ dofollow from random sites; zero nofollow across any niche |
| 4 | Anchor text distribution | Branded and generic anchors dominate; exact-match is a small fraction | Exact-match anchors form 10%+ of total; all anchors identical |
| 5 | Topical relevance of linking pages | Most links come from pages on related topics | Majority from unrelated niches or content farms |
| 6 | Link placement | Primarily in-content, editorial links | Concentrated in site-wide footers, sidebars, or boilerplate |
| 7 | Referring domain authority and traffic | Linking domains have real audiences; some traffic visible in Ahrefs/Semrush | Domains with zero organic traffic and no social presence |
| 8 | Acquisition velocity and pattern | Steady, gradual growth over time | Sudden spike with no corresponding PR or content event |
| 9 | Spammy TLD, PBN, and link-farm footprints | No meaningful cluster of .xyz, .loan, or other common spam TLDs | Many links share identical templates, hosting, or themes |
| 10 | Geographic and language relevance | Predominantly matches your target market | Mostly foreign-language or irrelevant-geography domains |
| 11 | Link scheme and paid-link red flags | No footprints of payment, reciprocation, or coordinated placement | Identical anchor and placement patterns across many domains |
| 12 | Disavow threshold assessment | Profile is clean; disavow reserved for manual action or provable link schemes | Over-reliance on third-party toxicity scores to trigger disavow |
Check 1: De-Duplicate to Referring Domains Before Anything Else
Open your export and ignore the raw backlink count until you have grouped it by referring domain. Ten links from one domain count as one referring domain. According to Ahrefs research on referring domains and backlinks, there is a strong positive correlation between the number of unique referring domains and organic search traffic, while multiple links from the same domain carry diminishing returns.
What good looks like: No single domain accounts for an outsized share of your total link count, and the referring domain list spans a variety of industries and geographies relevant to your site.
Red flag: Three or four domains account for 80% or more of all links. This either signals an old reciprocal link arrangement, a site-wide footer link, or a link-building campaign that lacked diversity.
Ahrefs and Semrush both separate referring domains from total backlinks in their exports. The GSC links export shows “top linking sites” sorted by link count, which makes it easy to spot outlier domains.
Check 2: Verify Live vs. Lost Link Status
A link in your export that no longer returns a 200 status is not contributing to your profile. Common failure modes: the linking page was deleted (returns 404), the linking domain expired, or the page redirects to something unrelated to your site.
What good looks like: Most links return 200 on a fresh status check. A reasonable percentage of link loss is normal over time, particularly for older profiles.
Red flag: A large portion of your link export returns 404, soft 404, or redirects to unrelated domains. If you built a campaign two years ago and a third of those links are gone, that history matters for diagnosing current profile strength.
BacklinkTower’s per-link status checks run against your export directly in the browser, flagging each link’s current HTTP status, including whether it has gone nofollow since you last checked.
Check 3: Check the Attribute Mix (Dofollow, Nofollow, Sponsored, ugc)
Since September 2019, Google has treated all four link attributes (dofollow, nofollow, sponsored, ugc) as hints rather than directives for ranking purposes. The practical implication: a profile with a mix of attributed links looks natural. A profile where every single link is dofollow, across thousands of domains and every conceivable niche, often does not.
For a full explanation of how each attribute works and when to use them, see the guide to dofollow, nofollow, sponsored, and ugc attributes.
What good looks like: The mix reflects the reality of how the site acquired links. Niche editorial sites naturally skew dofollow. Sites with heavy forum or comment participation show ugc and nofollow. Sponsored placements carry rel=”sponsored.”
Red flag: 95%+ dofollow across a large profile with no pattern that explains it. Or sponsored links that lack the proper attribute, which Google’s spam policies identify as a violation.
GSC’s links report does not show link attributes. You need an Ahrefs or Semrush export, or a tool that classifies attributes directly from the live link, to run this check. BacklinkTower classifies each link’s attributes when you load an export, without requiring access to either paid platform.
Check 4: Analyze Anchor Text Distribution and Spot Exact-Match Over-Optimization
A natural anchor text profile is dominated by branded anchors (your site name), generic anchors (“click here,” “this article,” “source”), and naked URLs. Exact-match anchors, where the anchor is the precise keyword phrase you are trying to rank for, are typically a small single-digit percentage of a healthy profile.
No universally fixed threshold exists, and Google has not published one. The practical signal is relative: if exact-match anchors account for a noticeably large share of your total profile, or if the anchors are suspiciously uniform across many different domains, that is worth examining.
What good looks like: Branded anchors form the plurality. Exact-match anchors are present but not dominant. Anchor variety reflects that many different people linked to you independently.
Red flag: Exact-match anchors concentrated above 10% of your profile, or many different domains using the identical anchor phrase. Both are footprints of coordinated link building.
Sort your anchor export by anchor text and look for any single phrase that appears disproportionately. BacklinkTower’s anchor-text distribution view flags over-optimization patterns when you upload an export.
Check 5: Assess the Topical Relevance of Linking Pages
A link from a page about software development pointing to a software company is topically relevant. The same link pointing to a payday loan site is not. Google has long used context and topical signals to assess link quality, and SpamBrain, Google’s AI-powered spam detection system, has grown more capable of identifying links acquired through coordinated schemes regardless of anchor text.
What good looks like: The majority of linking pages discuss topics adjacent or related to your site’s subject matter. Some off-topic links are fine and expected in a natural profile.
Red flag: Most of your referring domains are in completely unrelated niches, or the linking pages appear to exist purely to host outbound links with no real content audience.
Open a random sample of 20 to 30 linking pages from your export and scan their content. Manual spot-checking here is more useful than any automated topicality score.
Check 6: Identify Link Placement (Editorial vs. Boilerplate)
Where a link appears on the linking page matters. An in-content editorial link, placed within the body of an article that is relevant to the topic, carries more signal than the same URL sitting in a footer, a sidebar, or a widget repeated across every page of a site.
What good looks like: Most links are in the body of pages, appearing in context with surrounding text that is relevant to your content.
Red flag: Significant clusters of site-wide links (the same link appearing on every page of a domain), footer links with commercial anchor text, or links embedded in boilerplate widgets. Site-wide links count as one referring domain regardless of how many pages they appear on, but their placement devalues their signal.
Ahrefs and Semrush exports include a “link type” or “placement” column. Check it. If a domain contributes 400 backlinks but they all come from a site-wide footer, that domain’s contribution is one referring-domain-worth of signal, not 400.
Check 7: Sanity-Check Referring Domain Authority and Traffic
Third-party authority metrics (Ahrefs Domain Rating, Moz Domain Authority, Semrush Authority Score) are useful directional signals, but they should never be the primary criterion for a disavow decision. A domain with a low DR that has real organic traffic, engaged readers, and topically relevant content is often more valuable than a high-DR domain that exists only to sell links.
What good looks like: Linking domains have some evidence of a real audience, even if modest. Check for any organic traffic in Ahrefs or Semrush site explorer. Completely traffic-free domains with no indexed content are more suspicious than low-authority domains with real visitors.
Red flag: Referring domains with zero detectable organic traffic, no social presence, and no indexed pages other than link-laden directories. This pattern often indicates link farms or private blog networks.
Check 8: Examine Acquisition Velocity and Pattern
A sudden spike of dozens or hundreds of new links within a short window, unconnected to any content launch, PR campaign, or viral event, looks artificial to both humans and algorithms. Google’s systems evaluate link velocity as a signal of manipulation.
What good looks like: Link acquisition grows gradually over time. Spikes are explainable by events: a product launch, a press mention, a piece of content that got shared widely.
Red flag: A large cluster of links acquired in a tight timeframe, particularly if they share similar anchor text, linking domain profile, or IP ranges. This is a classic link-building campaign footprint.
Look at your referring-domain acquisition timeline in Ahrefs (Site Explorer > Referring domains > new/lost chart) or the equivalent view in Semrush. Compare any spikes to your own content or PR history.
Check 9: Scan for Spammy TLDs, PBN Footprints, and Link Farms
Private blog networks (PBNs) and link farms often share recognizable signals: heavy use of certain TLDs (.xyz, .info, .loan, .click, .top are over-represented in spam), identical WordPress theme templates across many domains, shared hosting IP ranges, thin content with no clear topical focus, and recently registered domains.
What good looks like: The TLD distribution in your referring domains roughly mirrors what a site in your niche would naturally attract. .com, .org, .net, and relevant country-code TLDs are predominant.
Red flag: A cluster of referring domains sharing any of the above PBN signals. Even a handful of PBN links can create a pattern that SpamBrain or a manual reviewer notices during a link spam review.
Google’s December 2022 link spam update, powered by SpamBrain, was explicitly designed to “neutralize the impact of unnatural links on search results” and the system has continued expanding its detection capabilities since.
Check 10: Check Geographic and Language Relevance
If your site targets English-speaking users in the US, UK, or Australia, the bulk of your link profile should come from English-language domains. A profile where most referring domains are in languages or geographies with no relation to your market is a mild but real signal worth noting.
What good looks like: The geographic and language distribution of linking domains broadly matches your target audience.
Red flag: Large clusters of links from domains in foreign-language markets you have never targeted, particularly if those links appeared in a short window. This is a common pattern in purchased link packages.
This check is rarely a standalone disavow trigger, but it can corroborate other signals when you are trying to determine whether a batch of links is organic or purchased.
Check 11: Look for Link Scheme and Paid-Link Red Flags
Google’s spam policies define link schemes as practices that manipulate links to or from a site primarily to manipulate rankings. Prohibited practices include buying or selling links that pass PageRank, excessive link exchanges, using automated programs to build links at scale, and guest post networks where the primary purpose is link acquisition rather than audience value.
What good looks like: The links in your profile have a plausible organic explanation. Editorial coverage, content partnerships, trade directories, citations in industry publications.
Red flag: Patterns of payment (anchors that look purchased, domains that host large volumes of outbound links to unrelated sites), mutual-link clusters among competitive domains, or footprints of automated link placement such as comment spam, forum profile links, or widget embeds at scale.
If you can identify links in your export that you knowingly acquired through schemes that violate these policies, document them. Those are legitimate candidates for outreach removal and, if removal fails, a disavow file.
Check 12: Decide Whether Anything Rises to the Level of a Disavow Decision
The disavow tool is a last resort, not a routine cleanup step. Google’s own guidance states that “most sites will not need to use this tool” and warns that “if used incorrectly, this feature can potentially harm your site’s performance in Google Search results.” For full context on when the tool is and is not appropriate, read the guide to whether to disavow these links.
The decision threshold is specific: disavow when you have a considerable number of provably artificial or spammy links AND there is a manual action in place, or you have strong reason to believe a manual action is imminent, AND you have already attempted to remove the links by contacting site owners directly.
What good looks like: You finish the audit and find no meaningful concentration of links that map to Google’s spam policies. Nothing to disavow.
Red flag: You have a confirmed manual action for unnatural links, or you can identify clear link-scheme links (paid, PBN, coordinated) that you are unable to remove through outreach.
Two common mistakes to avoid:
- Treating a third-party toxicity score (Semrush’s 0-100 score, Moz’s spam score) as an automatic disavow trigger. These scores are directional signals that require human review, not automated verdicts. Semrush explicitly states that human review is required before taking action on any flagged link.
- Disavowing links out of general anxiety about profile quality when no manual action exists and no clear link-scheme footprint is present. Over-disavowing removes link equity you have legitimately earned.
What to Do With What You Found
Run through all 12 checks and sort your findings into three groups:
No action needed: Links that pass checks 1 through 11 without significant red flags. Document the audit results and schedule your next review.
Monitor or investigate further: Links that raised yellow flags on topical relevance, velocity, or geographic distribution but do not clearly indicate manipulation. Keep them on a watch list for your next audit cycle.
Outreach and potential disavow: Links with clear link-scheme footprints. Start with outreach to the linking site to request removal. If outreach fails after documented attempts, add the domains to a disavow file. Build the disavow file methodically: domain-level entries where the whole domain is problematic, URL-level entries where only specific pages are the issue.
For active monitoring, scheduled re-checks that alert you the moment a link drops, goes nofollow, or changes anchor text catch profile changes before they compound. BacklinkTower’s monitoring layer runs these checks on a scheduled basis against your verified link list.
Frequently Asked Questions
How often should I run a backlink audit?
For most sites, a full 12-check audit once per quarter is sufficient. If your site has been through a link-building campaign in the past year, a Google spam update, or a ranking drop you cannot explain by content changes, run one immediately regardless of schedule. Sites under active link building should audit monthly.
What backlink export should I start from?
Start from whatever export covers the most of your real link profile. Ahrefs and Semrush indexes are larger and more detailed than GSC’s links report, which does not show link attributes and is not a complete crawl. GSC is useful for a quick sanity check and for confirming that Google has seen specific links, but for a full 12-check audit, use Ahrefs or Semrush as the primary export source. If you only have GSC, it is a valid starting point for checks 1, 2, 7, and 8.
Does a backlink audit help if I have no manual action?
Yes. Most backlink audits are not triggered by a manual action. They are routine profiling work: understanding what you have, catching link loss early, identifying anchor over-optimization before it becomes a problem, and making sure your monitoring covers the links that are actually contributing to your rankings. A manual action is a lagging indicator; a regular audit is the leading indicator.
Do I need a paid tool to run these checks?
Not for all of them. GSC’s links report is free and covers checks 1 and 8. Manual spot-checking covers checks 5 and 6 without any tool. Checks 3 (attributes), 4 (anchor distribution), and 7 (domain traffic) require either a paid Ahrefs or Semrush subscription, or a tool that classifies your existing export without requiring a separate index subscription. BacklinkTower works from an export you bring and handles attribute classification and anchor distribution without a separate index subscription.
How many toxic links are too many before I should disavow?
There is no numeric threshold. The disavow decision is qualitative: do the links in question clearly map to practices Google defines as link schemes, and have you tried to remove them through outreach first? A handful of clearly paid links with no other red flags in the profile rarely triggers a manual action on its own. A large cluster of PBN or link-farm links with matching footprints is a stronger candidate. When in doubt, refer to Google’s guidance: the disavow tool exists for sites with a considerable number of artificial links that have caused or are likely to cause a manual action, not for routine cleanup of a generally healthy profile.
Should I disavow links based on a tool’s toxicity score alone?
No. Toxicity scores from Semrush, Moz, and similar tools are directional signals that require manual review before any action. A high toxicity score on a link from a legitimate industry directory is a false positive. Automatically disavowing everything above a score threshold is a known way to remove link equity you have legitimately earned. Use the score to prioritize which links to investigate manually, then make the disavow decision based on what you find when you actually look at the linking page.
Leave a Reply