What Is an Algorithmic Penalty in SEO?
Most of what people call an algorithmic penalty is not a penalty at all. It is a system evaluating a site differently, plus other sites doing better. That distinction is not pedantry. It decides whether the right response is to undo something or to improve something. Those are opposite kinds of work.
The Term Is Doing Damage
Penalty implies punishment for a specific offence. Most situations described this way involve no offence at all. The word then sends businesses looking for something to undo when there is nothing to undo.
Why this is more than a quibble. There is a documented episode.
Before the manual actions report existed and made the distinction visible, site owners removed large numbers of links, submitted disavowals and spent considerable time fixing things when they either had no manual action at all or had one that was not affecting their rankings.
What that tells you. The confusion has a cost.
That was an entire industry acting on a diagnosis nobody had made, at real expense, to no purpose. It is the clearest evidence available that this is a terminology problem with a commercial consequence rather than a matter of using words precisely.
Why it still happens. The word arrives first.
A business whose rankings fell says penalised before anybody has established anything. From that moment the work is a search for something to remove. The search usually succeeds, because every site contains something that looks questionable if you go looking.
What this page does about it. Separates two situations.
They get the same name, need opposite responses and telling them apart is the whole exercise.
Nobody Reviewed Your Site
No person looked at your site. Nothing was announced. There is no report to check and no stated reason to read, which is precisely why people fill the gap with a theory.
What a manual action gives you. Certainty.
A person reviewed the site, a reason was stated and you were notified in a specific place. Whatever else that is, it is knowable. The whole of what is a Google manual action rests on that.
What this gives you instead. Silence.
Something changed, nobody told you, then no document anywhere explains what happened to your particular site. That absence is uncomfortable in a way that a stated reason is not.
Why the silence produces bad decisions. People do not tolerate a gap.
Faced with an unexplained fall, most businesses construct an explanation and then act on it. The explanation feels like knowledge because it accounts for the facts. It was assembled rather than discovered.
The discipline that helps. Say what you do not know.
Writing down which parts are established and which are guessed is unglamorous and it prevents most of the expensive mistakes in this subject.
Two Different Things Get Called This
Both get described as an algorithmic penalty and they are not the same event. Everything that follows depends on which one you have.
The first. A system detecting something the site is doing.
Automated detection of a practice, applied to that site because of what is on it. This is the closer of the two to a genuine penalty, since something specific is being responded to and that something can be identified.
The second. A broad update changing how quality is assessed.
Nothing about your site triggered it. The standard moved, every site was reassessed against the new one and yours came out differently. This is not a penalty in any sense. Calling it one is the source of most of the damage described in block one.
Why the second is much more common. It affects everybody at once.
Detection is targeted at sites doing a particular thing. An assessment change touches the entire index, so far more businesses experience the second while describing it as the first.
What they have in common. Only the silence.
Neither is announced and neither has a report. That shared absence is why they get merged. It is the only thing they share.
Undoing Against Improving
If something is being detected, the answer is to stop doing it and remove it. If assessment changed, there is nothing to remove and the answer is to become genuinely better. A penalty means undoing something. An update means becoming better.
Why these are opposite kinds of work. Not degrees of the same thing.
One is subtraction: find the thing, take it out, demonstrate it is gone. The other is addition: make the site more useful than it was and more useful than the alternatives. A team doing the first when the second is required is working hard in the wrong direction.
What the published position says. Improve and be reassessed.
Google's guidance on a normal ranking drop following an update is that no reconsideration request is needed, since there is no manual action to revoke. The route is to improve the site and allow it to be reassessed, which is the same distinction stated from the other side.
How the wasted year happens. Predictably.
A site falls after an update. Somebody calls it a penalty. Months go into removing content, disavowing links and stripping things out. The site becomes thinner rather than better, nothing improves, then the conclusion drawn is that the work was not thorough enough.
The link to the wider subject. Held in both places.
Our algorithm updates material states this from the update side and points here. Neither should be softened, because it decides how a business spends a year.
How To Tell Which You Have
There is no report to consult, so this is assembled from signals rather than looked up. Four questions narrow it considerably.
Did it coincide with a known update. The strongest single signal.
A fall lining up with a widely reported update, in an area many sites moved in, points away from anything your site did. Our algorithm updates material covers how to establish that timing.
Is it a section or everything. Scope narrows it fast.
Detection tends to affect what it detected, which is frequently one part of a site. An assessment change tends to move a site as a whole, since the standard applies everywhere.
Did competitors move too. The one people skip.
If several sites in your market changed position at the same time, something external happened. If only yours did, that points inwards.
Is there anything questionable in the history. Asked without embarrassment.
A site with a period of bought links or generated content behind it has a plausible candidate for detection. A site with no such history probably does not.
What to do with four answers. Weigh them.
This is a judgement rather than a calculation. It is worth making before spending anything, per how to recover from a Google penalty.
Recovery Is Not A Switch
Where a system reassesses a site periodically, improvement is not recognised the moment it is made. That delay is normal, it is the single most demoralising part of this situation and it is why people abandon correct work.
Why the delay exists. Reassessment is not continuous.
A site has to be revisited and evaluated again before anything can change. Until that happens, a genuinely improved site and an unimproved one look identical from outside.
What that does to a business. Erodes conviction.
Work is done, nothing changes, then the natural conclusion is that the work was wrong. The most common outcome is that a business reverses correct decisions shortly before they would have been recognised.
What we will not give you. A period.
No published source supports a timescale for reassessment and we do not invent one. Anybody quoting a number of weeks is describing their own experience of other sites rather than telling you anything about yours.
What to do with that instead. Decide in advance.
Agree before starting what improvement you are making and why it is right on its own merits. Work that was worth doing regardless of the fall does not need a reassessment to justify it. That framing survives the waiting.
There Is No Reconsideration Request
The review process exists for manual actions only. Submitting a request where no manual action exists achieves nothing. It is one of the commonest things a worried business tries.
How the design makes this clear. The function lives inside the report.
Google moved the request function inside the manual actions report, as a request review option, precisely because requests only work for manual actions rather than for algorithmic ones. If there is no action listed, there is nothing to request review of.
What the guidance says about an update. No request needed.
Google's position on a normal ranking drop caused by an update is that a reconsideration request is not required, since there is no manual action to revoke. Improve the site and allow it to be reassessed.
Why people try anyway. It feels like doing something.
A form to submit is enormously appealing when the alternative is improving a site slowly with no confirmation that it is working. The appeal is emotional rather than practical.
What the absence actually means. Nobody to convince.
Uncomfortable, though also freeing. There is no case to argue and no reviewer to satisfy. The only question is whether the site is genuinely better, which is a question you can answer yourself.
Most Drops Are Neither
Having spent a page separating two categories, the useful closing point is that most falls belong to neither of them. Seasonality, a competitor, a technical fault, a change the site made or a measurement error account for more declines than both categories combined.
Measurement error. The one nobody suspects.
A comparison against a longer month, a tracking fault, a filter somebody added. A meaningful share of reported drops turn out not to be drops. Our measuring performance material covers establishing whether one is real.
A change the site made. The one nobody remembers.
A rebuild, a migration, a plugin, a template, content removed by somebody being tidy. Frequently invisible to the person asking, because somebody else did it.
A competitor improving. Identical numbers, opposite cause.
Why this belongs at the end rather than the start. It would sound dismissive.
Told first, it reads as being waved away. Told after a full account of what a penalty actually is, it reads as the conclusion of an argument, which is what it is.
What to do first. The definite check.
Establish whether anything was reported, which takes a minute and rules out the only certain category, in what is a Google manual action. The full series is on the dangerous SEO practices guide.
Undo something?
Or become
better?
Those are opposite kinds of work and doing the wrong one costs a year. A review that establishes which situation you are actually in, before anybody removes anything. We will tell you plainly if we find nothing worth removing, because that is the usual answer.
What a risk review covers:
A risk review is a one-off piece of work rather than a monthly service, so it is scoped against the size of the site and how much history it carries.
Every guide.
One practice.
Black hat, grey hat, link risk, thin content, the spam policies, negative SEO, manual actions, algorithmic penalties, how to check for one and how to recover from one.
