How Long Do Rankings Take to Recover After a Migration?
There is no period we can give you. There is also a further answer most writing on this subject leaves out entirely: sometimes rankings do not fully recover. Both are uncomfortable. Both are true. Knowing them changes what a business should actually do.
The Uncomfortable Answer
Nobody can tell you how long recovery takes. The discomfort is precisely that there is no number to give. What can be described is the shape it usually follows and what would tell you it is not following it.
Why the question resists an answer. Too many variables.
Site size, how much changed, how well the mapping was done, how often the site is visited and what competitors did in the meantime. Those differ so widely between businesses that any single figure describes one site rather than a category of sites.
What a quoted figure usually is. Somebody's experience.
A supplier saying recovery takes a certain number of weeks is describing the sites they have worked on. That may be useful information about them. It is not a forecast about you. Treating it as one produces a deadline nobody can meet.
What is more useful than a period. A direction.
Whether performance is falling, flat or returning tells you considerably more than any calendar could. A site recovering slowly is behaving correctly, however far below its old level it currently sits. A site not moving at all is not behaving correctly. That distinction is available today rather than in six weeks, which makes it the more practical question by some margin.
The second uncomfortable part. Block six.
Recovery is not guaranteed. That is worth stating early rather than discovering late, after a year of patience. It is also the reason the earlier stages of a migration matter as much as they do. This page ends up arguing for work that happened months before anybody read it.
What Determines Recovery
The useful version of the answer is what recovery depends on. Five things do. Three were decided before launch, which is why this page is really an argument for the earlier ones.
How complete the mapping was. The largest factor.
Addresses that were mapped correctly carry their history forward. Addresses that were missed do not, nor ever will. No amount of waiting repairs a redirect that was never built in the first place.
How well destinations were matched. The second.
An old address pointing at a genuinely equivalent page has somewhere for its value to continue. One pointing at something approximate has less, which is a decision made months earlier showing up now, long after anybody remembers making it or could explain why.
How much else changed at the same time. The third.
A move that also rewrote content and redesigned everything is several changes being assessed at once, which takes longer and is considerably harder to read while it happens. Each additional change made at launch is paid for during recovery.
How often the site is visited. Not in your control.
A site crawled frequently is reassessed faster than one visited rarely. That reflects the site's existing standing rather than anything about the migration itself, which is why a smaller site can wait longer than a larger one for the same amount of change.
What competitors did. Also not in your control.
Positions vacated during a decline get taken by somebody else. Getting them back is a contest rather than a wait. The other party has no reason to concede them.
The Shape Of A Normal Recovery
A normal recovery has a recognisable form. Recognising it is more useful than timing it. Four phases, none of which has a duration attached.
The fall. Immediate and expected.
Performance drops shortly after launch, as the site is reassessed. Its presence tells you nothing at all about whether anything is wrong, which is why it is a poor thing to make decisions on.
The instability. The hardest phase to sit through.
Movement in both directions, pages appearing and disappearing, sections behaving inconsistently. This is processing happening rather than something failing. It is also where most panic occurs. Most of the damage described in block eight gets done here.
The turn. The moment that matters.
The decline stops deepening and begins to reverse. This is the signal worth waiting for. It is also the one people miss, because they are watching the level instead of the direction. It arrives well before performance is anywhere near where it was, which is precisely why it gets overlooked.
The gradual return. Slow and uneven.
Different pages recovering at different times, some quickly and some not at all. That unevenness is normal rather than diagnostic, because a move is processed address by address rather than as one event, per monitoring after migration.
When It Is Not Recovering
Three patterns mean something is wrong rather than merely slow. Recognising them early is the difference between fixing a problem this week and waiting out a fault that was never going to resolve itself.
Still deepening. The wrong direction entirely.
A normal recovery stops falling and then turns. Something still declining well after launch is not following that shape. Continuing to wait is waiting for something that is not coming.
Completely flat. No movement in either direction.
Instability indicates processing. A perfectly level line at a reduced level can mean nothing is being processed at all, which is a different problem from slow recovery and needs an entirely different response. Waiting helps only when something is already under way.
One section stuck while the rest returns. A pattern.
The most tractable of the three, because patterns have causes and causes can be found. If the affected pages share an address type, a template or a section, that shared property is where the fault sits. It is also where the investigation starts, which makes this the best of the three problems to have.
What to do with any of them. Investigate rather than wait.
Starting with what the site tells crawlers and whether the redirects behave, since those are quick to check and account for most of what goes wrong. The diagnosis is in redirect errors and recovery after migration.
Partial Recovery Is Common
Most migrations end with some things back where they were and some things not. That is the ordinary result rather than a failed one. Expecting it in advance prevents a great deal of unnecessary alarm and several unnecessary decisions.
What tends to return. The clearly mapped.
Pages with a direct equivalent, a clean redirect and content that survived intact. Where nothing was lost in translation, there is nothing standing in the way of the page being understood as the same page it was.
What tends not to. The approximate.
Pages mapped to something close rather than equivalent, pages that lost content in the rebuild and pages whose position within the site changed even though they still exist. The third of those is the one nobody looks for, because the page is present, correct and reachable.
Why the pattern is informative. It points backwards.
What did not recover usually identifies exactly what was compromised during the project. That is a list of things to fix rather than a mystery to endure, which makes partial recovery considerably more useful than it feels at the time.
What to do with it. Work the list.
Take the pages that have not returned, establish what is different about them and address whatever that difference turns out to be. That is ordinary work rather than recovery. It is available immediately, requires nobody's permission and is considerably more productive than watching a chart every morning.
Sometimes They Do Not Fully Recover
Full recovery is not guaranteed. Some migrations end permanently below where they started. Almost nothing written on this subject says so, because it is commercially inconvenient for everybody selling migrations.
Why it happens. Three ordinary reasons.
Links were lost when addresses stopped working and cannot be recreated, because somebody else owns them and made that decision years ago. Competitors took positions during the decline and are under no obligation to return them. Else content was lost in the rebuild and nobody noticed, so the pages genuinely are weaker than the ones they replaced.
What none of those is. A waiting problem.
Each is a change in the underlying situation rather than a delay in processing. More patience does not address any of them, which is why recognising them matters. Waiting quietly is the worst available response to all three.
Why we say it. It changes decisions.
A business told recovery is guaranteed waits indefinitely and does nothing. A business told it might not be starts rebuilding from where it actually is, which is the useful response and is available months earlier than the alternative.
What follows from it. The argument for everything earlier.
If full recovery were certain, the audit, the mapping and the redirects would barely matter, because everything would come back eventually regardless. They matter precisely because it is not certain, which makes this the most important sentence in the cluster rather than the most discouraging one.
What Actually Helps
Four things genuinely assist recovery. None of them is dramatic, all are available today and all sit within your own control, which is a considerably more useful position than waiting for something to happen.
Fixing broken redirects. The largest by far.
Any old address returning an error rather than a redirect is losing whatever it held. This is the single highest return action after a migration and it requires no judgement, only checking.
Correcting approximate mappings. The second.
Addresses pointing somewhere loosely related can be repointed at genuine equivalents. Where no equivalent exists, one can sometimes be created, which is a legitimate use of the time spent waiting.
Restoring lost content. The one people forget.
Pages shortened during the rebuild can be restored from the pre-migration record, assuming somebody thought to take one. A page that lost half its content is weaker for an entirely fixable reason rather than a mysterious one.
Updating important external links. Slow and real.
Asking the sites behind your most valuable links to point at the new address removes a dependency on the redirect and reaches the destination directly. Use the priority list from the audit rather than contacting everybody. Expect a modest success rate on a small number of valuable links.
What Does Not Help
Four things people do during recovery that make it slower, harder to read or both. Each is done from anxiety rather than ignorance. Each looks like responsible action at the time, which is why naming them is worth more than warning against them.
Changing things repeatedly. The commonest.
Each change restarts the reading and adds another candidate explanation. A few weeks of well-meant adjustments produces a site where nothing can be attributed to anything, at which point even a competent diagnosis has nothing to work with.
Rewriting content that was working. The counterproductive one.
Pages that lost position are frequently rewritten in an attempt to improve them, which feels productive. If the content was performing before the migration, the content is not the problem. Changing it removes the last thing that was known to work.
Undoing the migration. Rarely the answer.
Reversing means a second migration carrying every risk the first one had, applied to a site that has not finished settling from the first. It is not undoing anything. It is moving again, which should be recognised rather than described as going back.
Buying activity to compensate. The expensive one.
Spending on links or advertising to cover a decline treats the symptom, costs money and makes the underlying recovery harder to measure. It is worth doing only as a deliberate commercial decision, taken calmly, rather than as a reaction to a chart.
When To Accept The New Level
At some point, waiting for recovery costs more than working from where you are. That is a business decision rather than a search one. Nobody makes it, because nobody frames it as a decision.
What waiting actually costs. Attention.
Every week spent watching is a week not spent on anything else. The cost is not the waiting itself, which is free. It is everything the business did not do because it was still waiting for an answer.
What the decision point looks like. Three conditions together.
Performance has stabilised, the fixable things in block seven have been fixed and further movement has stopped in both directions. At that point what remains is the new baseline rather than a temporary position waiting to correct itself.
What accepting it means. Not giving up.
It means treating current performance as the starting point for ordinary work rather than as a deficit waiting to close on its own. The work is the same work either way. The framing is the only thing that changes. It also changes everything about how the business behaves, including what it is willing to commit to.
Why the framing matters. It unblocks everything.
A business waiting for recovery postpones decisions, budgets and campaigns until the numbers look right again. A business working from where it is makes those decisions now, which is the difference between a lost quarter and a lost year.
Sometimes
they do not
fully return.
Direction read rather than a calendar watched, the three patterns that mean stuck separated from the ones that mean slow, broken redirects fixed because that is the highest return available, nothing rewritten that was working before the move, with the new level accepted once the fixable things are fixed rather than waited out indefinitely.
What a migration engagement covers:
A migration is a one-off project rather than a monthly service, so it is scoped and quoted against the size of the site and how much is changing at once.
Every guide.
One practice.
What a migration is, the types, when to do it, planning, the checklist, auditing, mapping, redirects, internal links, tags, downtime, crawling, structured data, performance, notifying, monitoring, recovery and the mistakes.
