How Does an SEO Company Identify and Fix SEO Penalties?

A “penalty” is the word clients use. The reality an SEO company actually deals with is three distinct problems that look similar from the outside: a manual action a human reviewer at Google applied to your site, an algorithmic ranking drop tied to a spam-focused system update, or a toxic backlink profile that’s actively dragging your domain down even without a formal action attached to it. Each has a different diagnosis path, a different fix, and a different recovery timeline. Treating them as one problem (the most common mistake we see in DIY recovery attempts) is why so many “fixes” don’t actually move rankings back.

This is the violation-recovery side of SEO specifically. If your rankings dropped for reasons that have nothing to do with a penalty (content going stale, a competitor outranking you with a stronger page, a slow technical decline), that diagnostic process is different and covered in our guide to recovering lost rankings. And the ongoing discipline of monitoring and adapting to Google’s core updates before they cause damage is its own practice, covered in our guide to handling Google algorithm updates.

Step One: Confirm It’s Actually a Manual Action

Before doing anything else, an SEO company checks Google Search Console’s Manual Actions report (under Security & Manual Actions in the left-hand menu). This is the single fastest way to rule a manual action in or out, and skipping it is the most common reason recovery work starts in the wrong place.

If the report says “No issues detected,” the ranking problem is not a manual action. It’s either algorithmic or technical, and chasing a manual-action fix (writing a reconsideration request, for example) wastes time on a problem that doesn’t exist. If the report lists an active action, it will name the specific violation type, describe the affected URL pattern, and give Google’s own remediation guidance for that category. Google groups manual actions into named categories, and the fix is different for each one:

• Unnatural links to your site (link schemes pointing inbound, usually paid or manipulative)
• Unnatural links from your site (your own site is selling or passing links in a way that violates Google’s link spam policies)
• Thin content with little or no added value (scraped, auto-generated, or doorway pages)
• Cloaking and sneaky redirects (showing search engines different content than users see)
• Spammy structured data (markup that misrepresents the page’s actual content)
• Site reputation abuse (third-party content hosted on a reputable domain specifically to exploit that domain’s authority, a policy Google formalized in its March 2024 core update and spam policy expansion)

The category determines everything downstream. A site reputation abuse action means removing or noindexing third-party content sections, not touching backlinks. An unnatural links action means the opposite.

Step Two: Distinguish Algorithmic Drops From Manual Actions

When Search Console shows no manual action but rankings or traffic still dropped sharply, the working hypothesis shifts to an algorithmic cause. The diagnostic approach an SEO company uses here is straightforward correlation work:

1. Pull the exact date of the traffic or ranking drop from Search Console or analytics.
2. Cross-reference that date against Google’s confirmed spam and core update timeline.
3. If the drop lines up within a few days of a confirmed update, the cause is very likely algorithmic, and the fix is to identify which spam or quality signal the update targeted and address it directly rather than waiting it out.
4. If there’s no date correlation and no manual action, the problem is more likely technical (indexing, crawlability, a site migration that went wrong) and the investigation moves to a technical audit rather than a penalty-recovery process.

This is a meaningfully different system than the one most site owners assume exists. Google doesn’t really issue “algorithmic penalties” against individual sites the way it issues manual actions. What happens instead is that systems like SpamBrain, Google’s AI-based spam-prevention system, continuously evaluate sites against Google’s spam policies, and a site that trips those signals can lose rankings broadly the next time a relevant update reprocesses the web. SpamBrain was first named publicly in Google’s 2021 webspam report, and Google has since expanded its role into specific update categories, including the December 2022 link spam update, which used SpamBrain to detect both sites buying links and sites built specifically to pass outbound link value. Google does not publish a current daily-volume figure for how much spam SpamBrain processes; the most concrete recent disclosure is from the March 2024 core update announcement, where Google reported that the update reduced low-quality, unoriginal content in search results by 45 percent, exceeding its own internal 40 percent target. That’s the kind of figure worth citing, because it’s current, dated, and traceable to Google directly, unlike the recycled “40 billion spam pages a day” number that circulates across SEO blogs (that figure is real, but it’s from Google’s 2020 spam report and hasn’t been updated since, so presenting it as a description of what SpamBrain does “now” is misleading).

The practical implication for clients: an algorithmic drop tied to a named update (March 2024’s scaled content abuse and site reputation abuse policies are common current culprits) doesn’t get fixed by a single form submission the way a manual action does. It gets fixed by remediating the underlying issue and then waiting for Google’s systems to reprocess the site, which usually means the next relevant core or spam update.

Auditing for Toxic Backlink Profiles

Whether or not a manual action is present, a backlink audit is standard practice anytime a site’s organic visibility drops unexpectedly. An SEO company runs the site’s full link profile through dedicated link intelligence tools, most commonly Ahrefs’ Site Explorer (toxic link flags based on spam score and anchor text patterns), SEMrush’s Backlink Audit tool (which assigns a 0-100 toxicity score per domain), and Majestic’s Trust Flow versus Citation Flow comparison (a large gap between high citation volume and low trust signals that a link profile is built on quantity rather than legitimate authority).

The audit isn’t just counting flagged links. It’s separating three categories:

That last category matters more than most site owners expect. Google’s own guidance on the disavow tool is explicit that it should only be used when a site has a considerable volume of spammy, artificial, or low-quality links that have caused (or are likely to cause) a manual action, and that direct removal should be attempted first. Over-disavowing healthy links is a real way agencies do more damage during a “cleanup.”

Building and Submitting a Disavow File

When disavowing is warranted, the file itself follows a strict, documented format. It’s a plain .txt file, UTF-8 or 7-bit ASCII encoded, one entry per line, uploaded through Search Console’s disavow tool. A real disavow file looks like this:

# Disavow generated [date] following backlink audit
# Domain-level disavow: spam network identified in Ahrefs toxic flag report
domain:cheap-links-network.example
domain:expired-pbn-site.example

# URL-level disavow: single spammy page, not full-domain
http://low-quality-directory.example/listings/spam-page.html

# Domain-level disavow: negative SEO attack pattern, confirmed Nov 2025
domain:attack-source.example

Lines starting with # are comments Google ignores, which is why agencies use them to document why each entry was disavowed (useful both for the client record and, if a manual action follows, for the reconsideration request itself). After upload, Google has stated it takes roughly a few weeks for the disavow list to be incorporated as it recrawls and reprocesses the relevant pages, so this step alone isn’t a fast fix.

Filing a Reconsideration Request Correctly

If a manual action is confirmed, the fix has to be complete before a reconsideration request goes in. Google’s review process checks whether the violation has actually been resolved site-wide, not just on the sample URLs shown in the report. A request submitted before every affected page is fixed (or before a disavow file has had time to process, for link-based actions) typically gets rejected, which costs another full review cycle.

A reconsideration request that’s likely to succeed includes:

• The specific issue Google identified, in Google’s own terminology from the report
• What was changed, with enough specificity that a reviewer can verify it (URLs fixed, content removed, links disavowed)
• Evidence the fix is complete and ongoing, not a one-time patch (for link issues, this usually means the disavow file plus documented outreach attempts)

How Long Recovery Actually Takes

Google does not publish a guaranteed turnaround time for any of these processes, and any source claiming an exact day count for a “2026 study” should be treated skeptically. What Google and credible industry reporting do say:

The honest summary most practitioners give clients: a clean manual action fix can resolve in under a month, a link-based manual action often takes longer because Google explicitly treats those reviews as more involved, and an algorithmic drop tied to a spam or quality update doesn’t fully resolve until Google’s next relevant systems pass, which in practice means anywhere from a few months to closer to a year for severe cases. This is general practitioner consensus built from watching real recoveries play out, not a single dated study, and any agency quoting you a precise day-and-month range from an unnamed 2026 “report” is citation-laundering a guess.

Monitoring Recovery, Not Just Waiting for It

Recovery isn’t a single event to wait for. Once fixes are submitted (reconsideration request filed, disavow uploaded, or remediation complete ahead of the next update cycle), an SEO company tracks a specific set of signals rather than just refreshing rankings daily:

• Search Console’s Manual Actions report, checked weekly until it clears
• Indexed page count and crawl stats, to confirm Google is actually re-crawling the fixed or removed pages
• Keyword position tracking segmented by the affected URL pattern specifically, not site-wide averages, since a partial recovery can be masked by unrelated pages moving in the opposite direction
• Referring domain count and toxicity score trend in whichever link tool ran the original audit, to confirm the disavowed links are actually dropping out of the profile

How that recovery data rolls into a client’s broader reporting and the metrics used to judge whether a campaign overall is on track is its own framework, covered in our guide to measuring SEO campaign success.

The distinction worth holding onto through all of this: a manual action is a conversation with Google that ends when a human reviewer says the violation is resolved. An algorithmic drop is a systems problem that resolves on Google’s schedule, not yours, once the underlying cause is actually fixed. Confusing the two, or treating either one as a single quick form submission, is the most common reason penalty recovery efforts stall out.