The complication is structural, not procedural: Google’s manual actions system attaches the penalty to the site (the property in Search Console), not to whichever party actually did the work that triggered it. A reconsideration request that explains “our previous agency did this, we had no knowledge of it, we’ve since switched vendors” is providing context, but context isn’t remediation. Google’s own reconsideration request documentation is explicit that what reviewers look for is evidence that the violation has been fixed and, where applicable, that the site has taken steps to prevent recurrence. It doesn’t evaluate or weigh who was at fault. A site owner inherits full responsibility for anything published or linked under their domain, regardless of whether it was an in-house team, a freelancer, or an agency that produced the violation, and regardless of whether that agency is still under contract. This is worth stating plainly because it’s the single most common misunderstanding in these situations: there is no fault-based leniency tier in the manual actions process, and reconsideration requests that lead with attribution rather than remediation tend to get rejected or, worse, get a generic reply that doesn’t move the case forward because the reviewer had nothing concrete to verify.
Why attribution doesn’t help and remediation evidence does
Manual actions exist to flag a site’s presence in search results as unreliable given a detected violation of the spam policies, whether that’s unnatural links, thin/auto-generated content, cloaking, or another category shown in the Manual Actions report in Search Console. The review process when a reconsideration request comes in is essentially: has the specific violation identified in the report been addressed, and is there enough evidence to confirm it. A reviewer isn’t assessing intent or culpability, they’re assessing current state. If the manual action is for unnatural links (either to or from the site, most commonly a links scheme related to link-building the prior agency ran), the remediation Google expects is the same regardless of who built the links: identify the unnatural links, remove as many as can reasonably be removed by contacting the linking sites, and disavow the remainder using the disavow tool. If the manual action is for content-based violations (scaled content abuse, thin affiliate content, cloaking), remediation means removing or substantially rewriting the offending pages, or noindexing/removing them if they can’t be brought into compliance.
This creates a specific complication when the previous agency’s work is difficult to audit. If the agency built links across guest posts, PBNs, or paid placements without documentation, or produced hundreds of auto-generated pages without a clear content inventory, the current team’s first job isn’t writing the reconsideration request, it’s reconstructing what was actually done. A reconsideration request submitted before that inventory is complete tends to be vague (“we believe the issue has been addressed”) in a way reviewers can’t verify, which is functionally the same outcome as not having fixed anything. The practical consequence is that switching agencies doesn’t reset the clock or create a presumption of good faith; the new team has to do forensic work the old team should have documented, and the reconsideration request only goes in once that work produces something checkable, like an exported disavow file or a list of removed/changed URLs.
There’s also a secondary complication worth flagging: if the relationship with the previous agency is contentious (a dispute over payment, a lawsuit, a public falling-out), it can be tempting to use the reconsideration request as a venue to document that dispute for the record. Google’s documentation doesn’t ask for this and reviewers aren’t positioned to adjudicate a vendor dispute; including it doesn’t help the request and can make the actual remediation evidence harder to find in a long submission. The request should be about the site’s current state, not about who’s to blame for its prior state.
What to do about it
Start with a full audit of what triggered the manual action, using the specific category named in the Manual Actions report in Search Console as the starting point rather than assuming you know what the prior agency did. For link-based manual actions, pull backlink data from Search Console’s Links report and a third-party link index to build as complete a picture as possible of what was built, since the prior agency’s own records may be incomplete, inaccessible, or simply gone if the contract ended badly. Attempt outreach to remove links where feasible, document the attempts, and disavow at the domain level for anything that can’t be removed, since domain-level disavow is generally the more defensible approach when the link’s context and quality can’t be fully verified.
For content-based manual actions, inventory the affected page types (Search Console typically indicates the type of violation, and sometimes example URLs), and go through each template or content class the prior agency created to decide whether it can be rewritten to add genuine value or needs to be removed. Auto-generated pages built purely for programmatic scale without unique informational value are the common trigger here, and partial fixes (removing some but leaving similar pages live) tend not to satisfy reviewers, since the underlying pattern is still detectable across the rest of the site.
Once remediation is complete, the reconsideration request itself should be direct and specific: state what the violation was, what was found in the audit, exactly what was done to fix it (with counts, dates, and links to evidence where possible, such as a disavow file submission date or a list of removed URLs), and what process changes prevent recurrence, such as ending the relationship with the prior agency and instituting content or link-building review before publication. Mentioning that the work was done by a previous agency is fine as one line of context, but it should never be the substance of the request. The site’s current compliance is the only thing being reviewed, and the request succeeds or fails on the strength of that evidence alone.
