What Not to Do in the First Week After a Traffic Drop: Six Costly Reactions

Most of the expensive damage in algorithm recovery is self-inflicted, and almost all of it happens in the first week. Not because people are careless, but because doing nothing feels like negligence when traffic is falling and somebody is asking what you are doing about it.

The uncomfortable truth is that the correct first week is mostly measurement. Here is what to avoid, roughly in order of how much harm I have watched each one cause.

Do not upload a disavow file

This is the worst panic reaction available, because it is slow to take effect, slow to reverse, and capable of causing real harm on its own.

A disavow file tells Google to ignore links. If you include domains that were passing legitimate value, you have removed that value, and the effect appears gradually over weeks as pages are recrawled. Undoing it takes the same weeks again. I have seen a site lose a further twenty percent from a disavow uploaded in week one against a drop that turned out to be a canonical fault.

If you have already uploaded one and traffic fell afterwards, it is reversible: traffic dropped after uploading a disavow file.

Since 2022, most link problems are neutralisation rather than penalties: Google stops counting the links rather than deducting anything. Disavowing links that are already being ignored achieves nothing. The genuine use cases are narrow and they are covered in the backlink audit and disavow guide.

Do not mass-delete pages

Pruning is a legitimate recovery technique and a genuinely bad first reaction. The difference is that pruning done properly starts with knowing which pages carry links, internal links and residual rankings, and that analysis takes longer than a week.

Deleting in a panic removes pages that were holding the site up alongside the ones dragging it down, and it is not reversible in any meaningful sense once the URLs have 404ed and been dropped from the index.

If you have already deleted and traffic fell, much of it is recoverable but the link part has a deadline: traffic dropped after I deleted pages.

If you are going to prune eventually, consolidate rather than delete: merge overlapping pages into one stronger page and redirect the others. That preserves what they earned. The method is in the content pruning strategy.

Do not redesign or migrate

Shipping a redesign into an active decline is the single most effective way to make a problem permanently undiagnosable.

You will never know afterwards which part of the loss was the original cause and which was the redesign, because both moved at once and the before-and-after comparison no longer isolates anything. And redesigns cause their own losses, as covered in traffic dropped after a website redesign.

If a redesign is already scheduled and cannot be moved, at minimum record a full baseline first: Search Console export, current URL list, current indexed counts.

Do not change several things at once

This is the most common version of the mistake and the least dramatic, which is why it goes unchallenged.

A team rewrites twenty pages, adds author bios, changes the internal linking, updates the titles and improves page speed, all in the same fortnight. Three months later something recovers. Nobody can say what worked, so nobody knows what to keep doing or what to apply to the next site.

Worse, if it gets worse instead, nobody can say which change caused that either. Sequence the work, record what changed on which date, and leave enough time between changes for a crawl cycle.

Do not act mid-rollout

Core updates run twelve to twenty-four days and rankings move in both directions throughout. Sites regularly recover part of a loss before an update finishes, with no intervention at all.

The June 2021 core update is the cleanest demonstration: Google announced in advance that it was incomplete and that a second part would follow in July. Sites that panicked in June were reacting to an interim state that Google had explicitly said was not final.

Check whether a rollout is active with the update tracker before drawing any conclusion from a graph.

Do not submit a reconsideration request

Only relevant if you have a manual action, and if the Manual Actions report is empty there is nothing to reconsider. Submitting anyway wastes time and signals that the diagnosis has not been done.

Check the report first. It takes one minute and it is the single highest-value check in the whole first week, because a manual action comes with an explicit statement of the problem while an algorithmic demotion comes with nothing.

What to do instead

Day	Action	Measurement (what it gives you)
1	Export 16 months of Search Console performance data	Your baseline, before retention removes it
1	Check Manual Actions and the Pages report	Rules out the two fastest causes
2	Find the exact break date from daily data	Tells you what to match against
3	Establish scope: site-wide, one template, or a few pages	Separates quality, technical and competitive causes
4 to 7	Check technical basics and write down the hypothesis	A testable theory rather than a list of guesses

That is the whole of a good first week. The 48 hour checklist covers the first part in more detail, and the four causes triage covers the diagnosis itself.

One last thing, and it is the hardest to sell internally. Doing nothing to the site for a week is an active decision that protects your ability to understand what happens next. It is not inaction, and it is worth saying that out loud to whoever is asking for progress.

Frequently asked questions

What should I not do after a traffic drop?

Do not upload a disavow file, mass-delete pages, ship a redesign, submit a reconsideration request without a manual action, or change several things at once. Each destroys your ability to attribute what happens next.

Should I wait before making changes after a Google update?

Yes, until the rollout completes. Core updates run twelve to twenty-four days and rankings move both ways throughout, so sites sometimes recover part of a loss with no intervention.

Can a disavow file make things worse?

Yes. Including domains that were passing legitimate value removes that value, the effect appears gradually as pages are recrawled, and reversing it takes the same weeks again.

Is deleting old content a good first response?

No. Pruning is legitimate later, but it requires knowing which pages hold links, internal links and residual rankings. Deleting in a panic removes pages that were helping, and once URLs 404 and drop from the index it is not meaningfully reversible.

Holding that line internally is its own skill, and it is easier with a structure: how to explain a traffic drop to your boss or client covers what to lead with and what not to promise.