Three confirmed Google updates landed inside roughly five weeks this spring. Which means most sites whose traffic moved during that window genuinely cannot tell which one caused it — and the appropriate response differs substantially depending on the answer.
What actually ran
The timeline first, because a lot of coverage conflates these and one of them wasn't a Search update at all.
| Update | Dates | Duration |
|---|---|---|
| February Discover core update | 5–27 February | ~22 days |
| March spam update | 24–25 March | Under 20 hours |
| March core update | 27 March – 8 April | ~12 days |
| May core update | 21 May – 2 June | ~11 days, 21 hours |
For context, the December 2025 core update ran 11–29 December, about 18 days.
Two things are worth pulling out. The February update was Google's first affecting Discover alone rather than Search — initially for English-language users in the United States, with stated intent to expand. Sites that lost Discover traffic in February and assumed a Search problem were diagnosing the wrong system.
And the May core update reportedly rolled out in three distinct volatility spikes rather than one smooth curve, with practitioners describing it as producing heavier movement than the already-turbulent March update, across verticals and countries. Google issued no new guidance alongside it.
"Recovery" is the wrong frame
The most consequential correction in this article, and it's not a semantic quibble.
Core updates are recalibrations of how quality and relevance get assessed across all queries at once. They are not penalties applied to particular sites. Google doesn't publish recovery steps precisely because there's nothing you did wrong to reverse.
A site that dropped wasn't punished. It was reassessed relative to everything else — and either something else got assessed more favourably, or the weighting of a signal it depended on changed.
The distinction that changes behaviour A penalty is something done to you that can be undone. A recalibration is a change in how everyone is measured. Only one of those has a fix, and it isn't the one people go looking for.
The practical cost of the wrong frame is real. Teams treating a drop as a penalty respond with frantic deletion, restructuring and technical fiddling — and Google explicitly advises treating deletion as a last resort rather than a cleanup ritual. Teams treating it as a reassessment ask a more useful question: what did the reassessment find lacking?
The attribution problem
The practical difficulty nobody's coverage solves for you.
A Discover update completed 27 February. A spam update ran 24–25 March. A core update began 27 March. Against a background of continuous unannounced adjustment that Google acknowledges runs between confirmed updates.
So a site whose traffic changed anywhere from late February through April has at least three candidate causes plus ambient noise — and the fixes are entirely different. A spam update response is about identifying policy violations. A core update response is about content quality relative to competitors. A Discover response concerns a feed you may not even be optimising for.
One useful sanity check that costs nothing: if your email engagement, direct visits and social traffic all declined in the same period, the algorithm probably isn't the root cause. Something else changed, and you'd be optimising the wrong thing.
Nightwatch
Accurate daily rank tracking tool with beautiful visual reporting dashboards
Best for: Daily Rank Tracking
What the recovery data actually is
Worth being direct, because the honest answer to the title's question is uncomfortable.
Google publishes no recovery statistics. Every figure in circulation — percentages of sites that recovered, proportions of traffic regained, industry-by-industry winner and loser tables — comes from third-party tracking tools or individual agency observation, using their own samples and their own methods.
That doesn't make them worthless. Tracking tools see real movement, and experienced practitioners observe real patterns. But claims like "sites regained 60–70% of lost traffic within six to eight weeks" are practitioner anecdotes, and they get repeated downstream without their original caveats until they read as established fact.
The reliable signal is agreement across independent sources on direction, not any specific number. On that basis, the patterns reported consistently for 2026 were:
- Topical depth outperformed breadth. Sites with fewer, deeply-researched pieces on a narrow subject reportedly did better than high-volume operations chasing every keyword variant.
- First-hand experience gained ground — real examples, original testing, case studies — against generic rewrites and scaled summaries.
- Recalibration rather than targeting. Both core updates were described as adjusting how existing signals are weighted, not as introducing new penalties for specific tactics.
Treat even those as directional. They're also, conveniently, what commentators said after most recent core updates — which is either evidence of a stable direction or evidence of a stable narrative, and it isn't possible to tell from outside.
What Google actually says about timing
Because this is where expectations go wrong most often.
Google's position: some changes take effect within days, others need weeks or months, because its systems need to confirm an improvement is sustained. Meaningful recovery frequently requires a subsequent broad core update.
But there's a qualifier that gets dropped from most summaries and matters practically. Google also states that smaller unannounced adjustments run continuously between confirmed updates, and those can produce improvement in the interim. So the common advice that you must wait for the next named update is more pessimistic than Google's own guidance.
Two procedural points from Google worth following exactly:
- Wait at least a full week after rollout completion before drawing conclusions from Search Console.
- Compare the post-update week against the week before the rollout started — not against the rollout period itself, which is unstable by definition.
Both exist to prevent the same failure: reading noise as signal and making changes in response to it. That's the same discipline that makes a holdout group valuable in any measurement work, as set out in refreshing content properly.
The frequency shift is the actual story
More consequential than any individual update, and under-discussed.
Four confirmed ranking updates in five months, three of them inside five weeks, against continuous unannounced adjustment. The mental model of an algorithm that changes a few times a year no longer describes what's happening — it's closer to a dial that's always turning, with occasional larger movements that get names.
Which changes what a sensible response looks like. If updates are frequent and overlapping, then:
Reacting to each one individually is a losing strategy. By the time you've diagnosed and responded, the next one has landed and confounded your result.
Maintenance beats recovery. A library reviewed continuously is in better shape when an update lands than one reviewed reactively afterwards — which is an argument for a standing content audit rhythm rather than an emergency one.
Structural work compounds across updates. Topical depth, genuine expertise, clean technical foundations and sound internal structure survive recalibration better than tactics tuned to one weighting — which is the case for the approach in building topical authority.
What to do, in order
- Date the change and map it to the confirmed update windows. Split Search from Discover.
- Rule out non-algorithmic causes — seasonality, tracking breakage, a site change of your own, a lost link, a competitor launch.
- Wait a week past rollout completion before analysing, and compare against the pre-rollout week.
- Diagnose by page type and category, not sitewide. Broad declines and concentrated ones have different explanations.
- Look at what now outranks you for the queries you lost. This is the most informative single exercise available, and it's frequently skipped in favour of auditing your own site in isolation.
- Prioritise your highest-value pages rather than working alphabetically through a list.
- Improve rather than delete. Deletion is a last resort per Google's guidance, and it forecloses recovery on pages that might have recovered.
- Check technical fundamentals in parallel, since crawl and delivery problems compound anything else — the diagnostics in log file analysis are useful here.
Step five deserves the emphasis. A core update is a relative reassessment, so the question isn't only "what's wrong with my page" but "what does the page now ranking above me have that mine doesn't." That reframes an unbounded quality audit into a specific, answerable comparison.
And if the pages that dropped are AI-assisted and thin on original substance, the honest diagnosis may be that the reassessment found what it was designed to find — which is a production problem rather than an SEO one, addressed in the AI content QA checklist.
A note on what happened locally
Briefly, since local search moved distinctly.
The March update produced reported local-specific effects that don't map neatly onto the general picture — profile completeness gaining weight, review recency overtaking raw counts, and heavier volatility in specific mid-March windows. Local businesses diagnosing a March drop against general core-update guidance may be looking at the wrong signals entirely, which is unpicked in why local rankings keep swinging.
Similarly, if your drop coincided with a site migration or a redirect change rather than an update window, that's a different investigation entirely — the checks in redirects and migrations apply.
If the practical position is that you've lost significant traffic, can't confidently date it, and the next update is approaching, that's a diagnostic job with a deadline — and it's where an SEO partner who has run this sequence before is considerably faster than working it out under pressure.
The short version
Four confirmed updates ran through mid-2026 — a Discover-only update in February, a spam update in March, and core updates in March and May — with three landing inside roughly five weeks, which makes attribution genuinely hard and matters because the fixes differ. "Recovery" is the wrong frame: core updates are recalibrations of how everyone is measured, not penalties applied to you, which is why Google publishes no recovery steps. Every recovery statistic you'll read comes from third-party tracking rather than Google, so trust directional agreement across sources and distrust specific percentages. Date your impact precisely, wait a week past rollout, compare against the pre-rollout week, and look at what now outranks you rather than auditing yourself in isolation. And note that Google says smaller adjustments run continuously — you don't necessarily have to wait for the next named update.
Lost traffic and not sure which update caused it?
We date the impact, find what actually changed, and rebuild before the next one lands.
Explore SEO Services →Frequently asked questions
What Google updates ran in 2026?
Four confirmed ranking updates appeared on Google's Search Status Dashboard through mid-2026. A Discover core update ran from 5 to 27 February and was notable as Google's first update affecting Discover alone rather than Search. A spam update ran across 24 to 25 March in under twenty hours. The March core update ran from 27 March to 8 April, roughly twelve days. The May core update ran from 21 May to 2 June, around eleven days and twenty-one hours. A further broad update was anticipated in the third quarter.
Is 'recovering' from a core update the right way to think about it?
Not really, and the framing causes practical harm. Core updates are recalibrations of how quality and relevance are assessed across all queries, not penalties applied to particular sites. Google does not publish recovery steps precisely because there is nothing you did wrong to reverse. A site that dropped was not punished; it was reassessed relative to everything else. Treating it as a penalty leads to frantic deletion and restructuring, when the productive response is improving what the reassessment found lacking.
How long does it take to see improvement after a core update?
Google's own guidance is that some changes take effect within days while others need weeks or months, because its systems first need to confirm an improvement is sustained. Meaningful recovery frequently requires a subsequent broad core update. However, Google also states that smaller unannounced adjustments run continuously between confirmed updates and can produce improvement in between, so waiting passively for the next named update is not the only option available.
Why is it hard to tell which update affected your site in 2026?
Because the updates overlapped. A Discover update completed on 27 February, a spam update ran on 24 and 25 March, and a core update began on 27 March — three confirmed changes inside roughly five weeks, against a background of continuous unannounced adjustment. A site whose traffic moved during that period cannot easily attribute the change to one cause, which matters because the appropriate response differs substantially between a spam update and a core update.
Can you trust published data about which sites recovered?
Treat it cautiously. Google publishes no recovery statistics, so every figure in circulation comes from third-party tracking tools or individual agency observation, using their own samples and methods. Claims such as a specific percentage of traffic regained within a set number of weeks are practitioner anecdotes rather than measured findings, and they are frequently repeated without their original caveats. The directional patterns reported across multiple independent sources are more reliable than any specific number.
