Disavowing links benchmarks: what good looks like

Last Tuesday a client of mine—let’s call him Marco—rushed me a disavow file he’d built in a panic. He’d been hit with a manual action, spent two days burning through Google’s toxicity score list, and submitted a file with 120 domains. Three weeks later his organic traffic was down 18% rather than up. He’d disavowed good referring domains alongside the spam, tanked his authority signals, and had to file a reconsideration request to undo the damage. We’re all guilty of treating disavowing like a numbers game, but benchmarks exist for a reason: so you know when enough is enough and when to keep digging.

Why most disavow attempts fail before they start

The vast majority of failed disavow campaigns share a single root cause: treating toxicity as a binary switch rather than a spectrum. I’ve seen sites disavow entire domains because a single referring page triggered a high-score flag, only to realize later that the domain hosted a legitimate editorial partnership on a subfolder. Google’s algorithm doesn’t care about your spreadsheets; it cares about relevance, anchor text distribution, and the genuine editorial value of the link. If you dump a batch of links without mapping out which pages are actually harming versus merely unrelated, you’re just deleting authority without fixing the penalty.

Another common blind spot is relying exclusively on automated scores. Tools will flag links based on metrics like domain rating, spam score, or historical penalty patterns, but those metrics don’t account for context. A link from a forum with a 0.2 spam score might be perfectly fine if it’s a niche community discussion, whereas a link from a seemingly authoritative news site with a 0.5 score could be part of a private blog network. I always tell clients to start with a manual audit of the top 20–30 flagged links, identify the pattern, and then scale the disavow from there.

Setting realistic benchmarks for your niche

Benchmarks are useless if they’re not anchored to your specific industry. A news site can absorb an order of magnitude more spammy links than a SaaS startup before Google intervenes, because news sites naturally attract diverse, unmoderated commentary. In our work, we’ve found that most recovery-focused disavows fall between 10 and 25 toxic referring domains, but the median drops to 5–10 for niches where editorial standards are tighter—think finance, health, or legal. If you’re outside those verticals, expect the number to creep higher simply because the baseline link profile is noisier.

The most useful benchmark I can offer is a ratio: aim to disavow no more than 5% of your total referring domains if you’re pursuing a recovery path. Any higher and you’re likely cleaning up a healthy profile rather than targeting genuine toxicity. For penalty recovery, the goal is to get that percentage under 2% while preserving the links that drive real referral traffic and authority signals. I usually pull the pre-disavow profile, run a toxicity audit, and calculate the ratio before we draft a single line into Google Search Console.

Step-by-step: Auditing and building a disavow file

1. Pull your current backlink profile from Google Search Console or your preferred crawler. Export the referring domains list and note the total count.

2. Run a toxicity audit using a tool you trust—many will auto-flag links based on historical penalty data, anchor text patterns, and domain authority. Don’t trust the score 100%; treat it as a sorting mechanism, not a verdict.

3. Manually review the top 20–30 flagged links. Open each one in a new tab, read the surrounding content, and ask: does this look like spam, or just unrelated? If it’s a forum post or a comment spam that you’d expect to see on a healthy site, leave it alone.

4. Group the remaining toxic links by domain. If multiple pages on the same root domain are flagged, you can often disavow the whole domain in one line. Be meticulous about verifying that no legitimate subpages would be caught in the crossfire.

5. Calculate your toxicity ratio by dividing the number of domains you’re about to disavow by your total referring domains. If it’s creeping past 5%, pause and reconsider which links you’re actually targeting.

6. Submit the file in Google Search Console and set a calendar reminder to check rankings and indexation after 30, 60, and 90 days. Recovery is rarely linear, and dropping the file is only step one.

Common mistakes that tank recovery

The most damaging mistake I see is disavowing by domain without auditing individual pages first. Marco’s case from the intro is a perfect example: he disavowed a regional news outlet’s entire domain because two inner pages were flagged, only to discover the homepage was a legitimate source of industry citations. That single move stripped him of roughly 15% of his referring domains, and it took three months of link-building to rebuild the lost authority.

Another frequent error is submitting a disavow file and then walking away. Google doesn’t reprocess these signals instantly, and your rankings may dip before they recover. I’ve had clients check their Search Console dashboard daily for the first two weeks, only to see movement at the 60-day mark. Set expectations with stakeholders early: this is a long game, not a quick fix. Also, avoid the temptation to disavow links that are merely low-quality but not actively harmful. If a link has been pointing to your site for years without a penalty, leaving it alone is usually the safer play.

Comparing disavow strategies: volume vs. precision

Some SEO shops claim you need to disavow hundreds of links to see movement, while others preach a “less is more” approach. The truth sits in the middle, and it hinges on the type of penalty you’re facing. For algorithmic downgrades caused by a broad link profile issue, a high-volume disavow—sometimes 50 to 100 domains—can reset the algorithm’s perception of your site’s trustworthiness. But for manual penalties tied to a specific toxic campaign, a precision disavow of 10 to 20 domains often suffices because the spam is concentrated rather than pervasive.

I’ve run both styles across different client accounts. The volume approach tends to work faster for sites with sprawling, uncontrolled link profiles, but it carries a higher risk of collateral damage. The precision approach is slower to show results, but it preserves more of your existing authority and is easier to defend if you need to walk back the decision. Choose the strategy based on your penalty type, not a generic rule of thumb.

Measuring what good looks like after disavow

Once the file is submitted, how do you know if you’ve hit the right benchmark? I look at three concrete signals. First, traffic trends: a healthy recovery shows a gradual upward slope over 60 to 90 days, not a jagged line of spikes and drops. Second, referring domain diversity: if your post-disavow profile has more unique domains than the pre-disavow one, you’ve likely cleaned up without sacrificing reach. Third, anchor text distribution: toxic profiles often have a few dominant, over-optimized anchors. After a clean disavow, that distribution should flatten toward a more natural mix of branded, generic, and partial-match anchors.

If any of those three metrics move in the wrong direction after 90 days, it’s a sign the disavow wasn’t specific enough—or that you disavowed too aggressively. I usually circle back with the client, re-audit the live profile, and adjust the next file. It’s iterative, and it demands patience, but the alternative is staying stuck in a penalty loop indefinitely.

The hidden math: calculating your personal toxicity threshold

Before you even open a disavow file, you need to understand that not all toxic links are created equal. The “toxicity” metric most tools spit out is usually a composite of domain authority, historical penalty patterns, and spam indicators. But what matters for your site is how Google actually weighs those factors in your specific niche.

I like to run what I call a “baseline toxicity audit” first. Pull your full referring domains report, tag every domain as either “clean,” “questionable,” or “toxic” based on your manual review of the top linking pages, then calculate the percentage. In our agency’s data set from the last 18 months, the average site sits between 3% and 8% toxic referring domains depending on vertical. If yours is sitting at 2% and you’re hit with a penalty, the problem isn’t the volume of toxic links—it’s the quality of the anchor text or the relevance of the linking domains. Conversely, if you’re at 15% and have never seen a ranking dip, Google may have already devalued those links naturally, and a mass disavow would be unnecessary risk.

The real insight here is that the 5% ratio I mentioned earlier is a ceiling, not a target. Some sites can safely disavow 7% of their profile and see no movement because Google has already discounted those links. Other sites, particularly newer domains or those in competitive verticals, might see a ranking boost from disavowing just 3% because the algorithm is still heavily weighting those signals. Do the math on your own profile before you start naming names.

Advanced: Cross-checking tools against manual reality

There’s a persistent myth that SEO tools can automatically generate a perfect disavow file. I’ve yet to see one that consistently nails it. Tools like Ahrefs, SEMrush, and the various “link risk” calculators are excellent for surfacing candidates, but they operate on proxy metrics—domain rating, URL rating, spam score—that don’t always align with how Google evaluates risk.

I run a simple cross-check every time: I take the top 50 auto-flagged links from whatever tool I’m using, manually open each one, and categorize them into three piles: “obviously toxic,” “probably fine,” and “unsure.” More often than not, the tool misses 20 to 30 percent of the genuinely harmful links because those links come from sources the algorithm isn’t trained to flag—think guest posts on niche industry blogs that are technically editorial but exist solely for link equity, or expired domains that have been repurposed as link farms.

Conversely, the tool will flag a handful of links that are actually harmless—perhaps a resource page on a government site that links out as a courtesy, or a university directory listing from five years ago. If you blindly disavow everything the tool flags, you’ll end up with a depleted profile, exactly the kind of authority loss Marco experienced. The magic happens in that “unsure” pile. Those are the links you invest the manual audit time in. If 10 links are in the unsure pile and 7 of them turn out to be clean after you read the content, you’ve just saved yourself from disavowing 7 good domains. That’s a net win.

When to walk away: recognizing false positives and stale data

Not every toxic link needs to come down. In fact, some of the most well-meaning disavow files I’ve reviewed have done more harm than good because the webmaster was chasing ghosts. There are three scenarios where I advise clients to leave the file empty.

First, stale links. If a link has been pointing to your site for three or four years and you’ve never received a warning or penalty related to it, the likelihood that it’s the cause of a sudden ranking drop is low. Google’s algorithm evolves; links that were risky in 2018 are often discounted today. Disavowing an old, devalued link is like pulling a tooth that’s already fallen out—you’ll just bleed authority.

Second, editorial links from high-trust domains that happen to have user-generated content sections. I saw a client disavow an entire university domain because a student forum page was flagged. The university’s main domain was a massive authority booster—academic citations, .edu backlinks, trusted references. Stripping that away for one noisy forum thread was a category-level error. Always check if the flagged page is a subdirectory or subdomain of a larger, legitimate entity before you pull the trigger on a domain-level disavow.

Third, links that are technically “spammy” but thematically relevant. There’s a difference between a link that violates Google’s guidelines and a link that’s just unrelated to your topic. A link from a low-quality directory in a completely different industry might have a high spam score, but if it’s not actively hurting your relevance signals and you have a healthy link profile elsewhere, let it be. The disavow file is for recovery, not for perfecting a profile that’s already functioning.

Real-talk conclusion

Disavowing links isn’t a magic wand, and there’s no universal number that guarantees a recovery. The benchmarks I’ve shared—10 to 25 domains for most niches, under 5% toxicity ratio, 2 to 6 months for results—are guardrails, not guarantees. What actually moves the needle is a disciplined audit process, the willingness to manually verify automated flags, and the patience to let Google reprocess the signals in its own time. If you’re staring at a spreadsheet wondering how many is enough, start with the ratio, audit the top flagged links by hand, and submit a file that targets the specific toxicity, not the total count.

If you’re tired of guessing whether your disavow file is doing more harm than good, OptiLink can help you flag potentially toxic links at scale, cross-check automated scores with manual reviews, and track your profile’s health over time so you’re not flying blind. It’s not a set-and-forget solution, but it does cut down the hours spent sorting through spreadsheets and lets you focus on the links that actually matter.

Keep reading

Playbook

Broken link building, step by step: find, fix, and pitch

A hundred random links can do less than five relevant ones. Here's what makes a backlink pass real ranking value in 2026 — and what makes it worthless or worse.

Sofia Meyer October 9, 20262 min read

Playbook

Competitor backlink analysis that finds real opportunities

A hundred random links can do less than five relevant ones. Here's what makes a backlink pass real ranking value in 2026 — and what makes it worthless or worse.

Sofia Meyer September 26, 20262 min read

Playbook

Digital PR: the most durable link building there is

A hundred random links can do less than five relevant ones. Here's what makes a backlink pass real ranking value in 2026 — and what makes it worthless or worse.

Sofia Meyer October 13, 20262 min read

Explainer

Dofollow vs nofollow in 2026: what still passes value

The link attributes changed, Google's treatment changed, and a lot of advice is out of date. Here's what dofollow, nofollow, sponsored and ugc actually do now.

The OptiLink Team October 10, 20262 min read