Manual actions and disavow

Lesson 42 of 50Module 8 · Measurement, risk and recovery5 min read

By Bill Hartzer · Hartzer Consulting · 25+ years in search, domains and expert witness work

The calibration that should change how you read your profile

This is the most important number on this site, so I want to state it precisely.

A profile built over twenty years, on which not one link was ever bought, contains 22,260 referring domains. Of those, 69.3% — 15,421 domains — sit at Trust Flow 0. No measurable trust at all. A further 18.1% are Trust Flow 1 to 10. Only 446 domains, 2.0% of the total, reach Trust Flow 41 or above, which works out at roughly 22 a year over two decades.

Run any commercial toxicity tool across that profile and it will flag many thousands of domains and produce a frightening report. Every one of those domains arrived without any action by the site owner. They are scrapers reproducing articles, aggregators listing feeds, abandoned directories, expired sites that changed hands, low-quality blogs copying paragraphs, and the general sediment that accumulates around anything that has been useful and online for a long time.

Junk links are the default state of publishing on the internet. They are not a symptom of anything, they are not evidence of manipulation, and their presence tells you nothing about your site except that it has existed for a while and someone found it worth copying. If you take one thing from this module, take this, because it is the single most common reason people damage their own profiles.

What a manual action actually is

A manual action is a specific thing with a specific location: a human reviewer at Google has assessed your site and applied a penalty, and it appears in the Manual Actions report inside Search Console. It has a stated type and a stated scope, either site-wide or partial. It comes with a reconsideration process.

Everything else is not a manual action. A traffic drop is not a manual action. A ranking decline after a core update is not a manual action. A competitor telling you that you have been penalized is not a manual action. If it is not in that report, it did not happen, and the number of profiles I have examined where somebody was treating an algorithmic ranking change as a penalty and preparing to disavow accordingly is remarkable.

The distinction matters because the responses are completely different. A manual action for unnatural links requires you to actually clean up the links in question and demonstrate that you did. An algorithmic decline requires you to improve the site, and disavowing links will not address it, because the ranking system was not responding to a penalty in the first place.

First step in every case, before any analysis: open Search Console and look. If the report is clean, put the disavow file away and go and find the real cause.

When disavow genuinely applies

The disavow tool exists for a narrow situation. In my experience it applies when all of the following are true:

• You have a confirmed manual action for unnatural links, visible in Search Console, or you are dealing with a documented history of link buying that you are actively cleaning up ahead of a review.
• You can identify the specific links that were manufactured, because you or a predecessor bought or arranged them, and you can list them rather than guess at them.
• You have tried to remove them at source and the attempt is documented. Removal is what the process actually asks for; disavow is what you fall back on for the ones you cannot reach.

There is a second, rarer case: a genuine negative link campaign, where somebody has pointed a large coordinated cluster of manufactured links at your site with commercial anchors, arriving together, from a network you can identify. This is much less common than people believe, and the bar is that you can demonstrate the coordination, not that you have noticed some bad-looking domains.

Outside those situations, the correct number of URLs in your disavow file is zero. Not a small file. Zero.

The test I apply is whether you can name the mechanism by which each entry got there. If the answer is that an agency bought them in a particular year through a particular vendor, that is a mechanism and a defensible entry. If the answer is that a tool scored them badly, that is not a mechanism, and the entry is a guess with permanent consequences attached to it.

Why most disavow files do harm

The disavow tool is one of the few instruments in SEO that lets you damage your own site quickly and quietly. Four mechanisms.

It is used on the wrong problem. Traffic falls after a core update, somebody runs a toxicity tool, thousands of domains get flagged because 69.3% of any mature profile looks like that, and the file goes up. The ranking cause is untouched and real links have been discarded.

Domain-level entries are blunt. Most files use the domain directive, which disavows every link from that domain forever. Sites change hands, improve and get cleaned up. A domain disavowed on a bad day stays disavowed.

Automated lists have no judgment. Tool-generated disavow files routinely include genuine relevant sites that happen to score badly on metrics: small niche blogs, forums, regional publications, personal sites. Those are frequently the most topically relevant links a site has.

Search engines already handle this. Ignoring unsolicited junk is a basic requirement of running a search engine, because otherwise anyone could damage a competitor for the price of a cheap link package. The system has to be robust to noise, and it largely is.

So the practical guidance is uncomfortable but simple. If you are not sure whether you need a disavow file, you do not need one. The people who genuinely need one know exactly which links they are disavowing and why, because they arranged them.

If you genuinely do need one

Working through a real cleanup, the sequence matters.

1. Confirm the manual action and read the stated scope. Site-wide and partial call for different amounts of work.
2. Build the list from knowledge, not from scores. Which links were arranged, by whom, when, through which vendor. Invoices, emails and agency records are your source, not a toxicity export.
3. Attempt removal at source and document every attempt. Dates, addresses, responses. The documentation is a substantial part of what a reconsideration review is assessing, because it shows intent.
4. Disavow only what you could not remove, at URL level where you can, using the domain directive only where the entire site was part of the arrangement.
5. Write the reconsideration request honestly. What happened, who did it, what you removed, what you disavowed, what has changed so it will not happen again. Reviewers read a great many of these and evasion is obvious.
6. Keep the file under version control and revisit it. A disavow file is a live document, and entries that were defensible three years ago may not be now.

One closing caution about consultants who lead with disavow. A profile with a lot of junk is a profile, and anyone who diagnoses toxicity from a metric distribution without knowing the site's history is describing the internet, not your problem.