Penalty-risk link audit with a defensible disavow file

A disavow file is the wrong first move in most of the cases that reach me. Of the 41 link investigations I have run since 2021, 17 ended with a file submitted to Google. The other 24 ended with a written finding that the decline had a different cause, a template change that stripped internal links, a canonical pointed at a filter URL, a category tree that quietly fell out of the index, and no file at all. Disavowing is not free: a domain you remove that was helping you is a loss you will never see and cannot reverse. So the first question this engagement answers is whether your link profile is the problem at all. What is decided, and by whom: I do the classification myself. No junior analyst, no offshore labelling team, no script that reads a third-party toxicity score and calls it a verdict. I have argued these decisions in front of a Google reviewer, in front of a CFO who had personally approved the invoices for the links in question, and in front of the agency that sold them. That is the standard the file is written to: every entry has to survive somebody hostile reading it. How a domain earns its place on the list: Referring domains are reconciled from four indexes, Search Console's link report, Ahrefs, Majestic and Bing Webmaster Tools, because none of them sees the whole graph and only one shows what Google admits to knowing. Each domain is then checked against a rubric you receive in full: is the linking page indexed; does the page have an editorial reason to exist; outbound link ratio; anchor distribution read across the whole profile rather than one link at a time; footprint overlap with other linking domains, meaning shared class C, shared registrant, identical theme, links published inside the same fortnight; and site history in the Wayback Machine, because a legitimate domain bought and repurposed in 2019 reads differently from one that was always spam. A domain goes on the list only when it fails two independent checks and I can write one sentence explaining why a reviewer would agree. On a normal profile that is three to nine per cent of referring domains. If somebody has told you that forty per cent of your links are toxic, you were sold a score, not a judgement. When the answer is inconvenient: Two versions of this come up. The first is that the links were bought and somebody still employed signed off. I put that in writing, with dates, because a reconsideration request that omits it gets rejected and costs you another month. The second is that the profile is clean and the cause is elsewhere. I stop there, hand over the finding and the evidence, and we close at the entry price. I would rather lose the larger fee than sell you a file you do not need. What arrives: An upload-ready disavow.txt with a comment line above every entry explaining the block. The full classified sheet, domain, verdict, evidence, source and date, built so the next specialist can check my work instead of starting again. A written finding of six to ten pages naming the cause or ruling links out. A removal list for the domains where asking is genuinely better than disavowing. A call in English or German with whoever has to sign the file off. Not included: New links. Content. Fixing the technical faults I identify, beyond naming them and telling you what they are costing. Payment to anyone demanding money to remove a link: I will not arrange it and I will advise you against it. A guarantee of recovery, because nobody controls Google's ranking systems, and a vendor promising restored traffic by a fixed date is telling you something about themselves. Not for you if: You want a large file to show a board that action was taken. You want it this week. You run link networks and want a footprint check before scaling. Or you want a monthly toxicity dashboard, which is a subscription, not an audit.