usdigitalmarketingindex.com SEO
How can you investigate an organic traffic decline beginning after a Google core update?
Updated November 15, 2026
By treating the update as one hypothesis among several, and testing each against the data before accepting any. A decline that begins near a core update’s dates may be caused by it. It may also be seasonal, the result of a change on the site, a tracking fault, or a shift on the results page, and the update may be a coincidence. Google’s own guidance frames the first step as a check, not a conclusion: if a drop seems to correlate with a core update, “use Search Console to determine whether you need to make a change” (Google’s documentation on core updates). The investigation confirms the timeline, finds which pages and queries moved and by how much, rules out the other explanations, and only then considers what, if anything, to change.
Confirming update timeline
Establish the dates from Google, not from commentary.
The guidance’s first instruction: “Confirm that the core update has finished rolling out”, by checking the Search Status Dashboard and noting the update’s start and end dates.
Then compare the right periods. The guidance says: “We recommend waiting at least a full week after a core update completes before analyzing your site in Search Console”, and suggests comparing that week with a week before the update started rolling out.
| Step | Detail |
|---|---|
| Record the update's start and end dates | From the Search Status Dashboard |
| Find when the site's decline began | Daily clicks and impressions in Search Console, by date |
| Compare | Did the decline begin within the rollout window, before it, or after it ended? |
| Wait | Do not analyze during the rollout; positions move while it is in progress |
| Choose comparison periods | A full week after completion against a full week before the start |
Timing findings and what they suggest:
| Finding | Reading |
|---|---|
| Decline began before the rollout started | Something else, at least in part |
| Decline began during the rollout | Consistent with the update; not proof |
| Decline began after the rollout ended | Less likely to be the update; look for site changes and other events |
| Gradual decline across months, with the update inside it | A longer trend |
| Sharp step on one day | An event: check deployments as well |
Consistency in time is necessary and not sufficient. Many things happen in any two-week window.
Separating affected page types
Find out what moved. The guidance says to “Review your top pages and queries” and assess how they ranked before and after.
Segment the comparison:
| Dimension | Question |
|---|---|
| Page type or template | Did guides fall while product pages held? |
| Topic or section | Is one subject area affected? |
| Query type | Informational, commercial, branded? |
| Individual top pages | Which lost the most clicks? |
| Device and country | Everywhere, or in one segment? |
| Search type | The guidance says to analyze search types separately: web, images, video, news |
Then size the change in position. The guidance distinguishes two cases, with its own examples.
| Case | Google's example | Google's advice |
|---|---|---|
| Small drop in position | From position 2 to 4 | "There's no need to take drastic action"; it recommends against changing content that is already performing well |
| Large drop in position | From position 4 to 29 | "Do a deeper assessment" |
Patterns and their meaning:
| Pattern | Suggests |
|---|---|
| Decline spread across page types and topics, with positions lower | Site-level reassessment, consistent with an update |
| Decline confined to one template | A technical or template cause; check releases |
| Decline confined to a few pages | Page-level causes: competition, content age, a results feature |
| Impressions down, position steady | Demand, not ranking |
| Clicks down, impressions and position steady | Click-through: a results page change |
| Branded steady, non-branded down | Competitive position changed |
Content and demand checks
Test the competing explanations.
| Hypothesis | Check |
|---|---|
| Seasonal demand | Same weeks last year; an outside measure of search interest |
| Site change | Change log and deployment history for the weeks around the decline |
| Technical regression | Indexing report, robots.txt, noindex, canonical, status codes, rendering on affected templates |
| Tracking fault | Search Console clicks against analytics sessions; if only analytics fell, it is measurement |
| Results page change | Search the main queries; record what appears above the site's result |
| Lost or changed pages | Were affected URLs removed, redirected, or rewritten? |
| Manual action | Manual Actions report |
| Competitor improvement | Which pages now rank above, and what they offer |
| Another update | The ranking updates list for other systems active in the period |
For pages with a large, sustained drop, the guidance points to self-assessment: read the affected pages against Google’s questions about helpful, reliable, people-first content, and look at what now ranks. It explains the idea with an analogy to a list of top restaurants, where those that move down “aren’t necessarily” bad and others have simply made the list, and it notes that other pages may be doing a better job of helping the searcher.
A content review for the affected set:
| Question | Evidence |
|---|---|
| Does the page fully answer the query it ranked for? | Compare with the pages now ranking |
| Is it accurate and current? | Check facts and dates |
| Does it show first-hand knowledge or original material? | Read it |
| Is it one of many near-identical pages on the site? | Site structure |
| Who wrote it, and is that evident? | The page |
| Was it written for readers, or to capture a query? | Honest reading |
Reassessment based on evidence
A core-update investigation worksheet. Every hypothesis starts as unconfirmed.
| # | Hypothesis | Status | Evidence for | Evidence against | Confidence |
|---|---|---|---|---|---|
| 1 | The core update lowered the site's positions | Unconfirmed | |||
| 2 | Seasonal fall in demand | Unconfirmed | |||
| 3 | A site release caused a technical regression | Unconfirmed | |||
| 4 | Analytics tracking changed | Unconfirmed | |||
| 5 | Results page features reduced clicks | Unconfirmed | |||
| 6 | Specific pages lost ground to better competitors | Unconfirmed | |||
| 7 | Content removed or changed on the site | Unconfirmed | |||
| 8 | A manual action applies | Unconfirmed | |||
| 9 | Several causes combined | Unconfirmed |
Statuses: unconfirmed, supported, ruled out. Move a hypothesis only when a specific check has been done and recorded.
Hypothesis 1 becomes “supported” when the decline began within the rollout, positions fell across a range of queries, the fall is not explained by the other rows, and the affected pages share characteristics. It cannot become “confirmed” in a strict sense. The attribution rests on timing and pattern, so the conclusion is an inference, and should be written as one.
What to do with the conclusion:
| Conclusion | Response |
|---|---|
| Small position changes | No drastic action, per the guidance |
| Technical or tracking cause found | Fix it; this is not an update matter |
| Seasonal | No action |
| Large, sustained, broad drop consistent with the update | Assess content against the self-assessment questions; improve where the assessment finds real weaknesses |
| Evidence mixed | Address what is established; keep observing |
Guidance on making changes: “Avoid doing ‘quick fix’ changes”, with the example of removing a page element because of something heard about SEO. Deleting content is described as a last resort, to be considered only where content cannot be salvaged.
On timing, the guidance is candid that improvement is not immediate: some changes can take effect in a few days, “but it could take several months for our systems to learn and confirm that the site as a whole is now producing helpful, reliable, people-first content”. It adds that after a few months without effect, that could mean waiting until the next core update.
Write the finding with its uncertainty. “Non-branded clicks to guide pages fell 30% between the comparison weeks. The decline began on the third day of the rollout. Seasonality, releases, and tracking were checked and ruled out. Positions for 40 tracked queries fell by an average of six places. This is consistent with the core update, and that attribution is an inference” is a defensible statement. “The core update hit us” is not a finding.
Do not act in the first days. Changing pages while positions are still moving makes later interpretation impossible.
A single update is one event. A site that has lost visibility across several years needs a broader plan, which is the subject of the guide to how to design an SEO recovery strategy for a website with several years of declining visibility.
The order of work: confirm the dates, wait at least a full week after completion, compare like weeks, find what moved, rule out the other causes, and state the update’s role as supported or not, never as proven.
Add your comment to this answer