Rankings changed, but we didn’t: how to diagnose the cause

Rankings can fall even when your website has not changed. Learn how to separate competitor movement, demand, SERP changes, indexing, updates and measurement noise before you edit anything.

“We changed nothing, so why did visibility move?” It is a reasonable question. A ranking report drops, someone spots a competitor above you and the instinct is to edit the page immediately.

That reaction can make a difficult situation harder to understand. Rankings are comparative: your page is assessed alongside other eligible pages, against a search result that can change and for a query whose demand and context can change. An unchanged website is not guaranteed to keep the same position.

The useful question is not simply, “What did Google do?” It is: which explanation best fits the evidence?

This guide sets out a practical way to investigate unexplained ranking movement. It covers competitor changes, search demand, search intent, SERP composition, indexing, confirmed search-system events and measurement noise. The aim is not to produce another list of algorithm theories. It is to help you decide whether to monitor, investigate or make a specific change.

Start with the possibility that nothing is wrong with the page

Google describes multiple automated ranking systems that evaluate pages and return results for particular queries. There is no single, static public rule that explains every movement. Google also documents that search results can vary according to context, including the query, location, language and device.

That means a position change does not, by itself, show that your page has become worse. Another page may have become more relevant to the query, a different result type may now be preferred or the measured result may simply differ from the one you saw previously.

The documentation provides general context, not the cause of an individual URL’s movement. Be wary of any analysis that jumps directly from “my ranking fell” to a named ranking factor.

In practice, treat the movement as a set of competing explanations:

• another eligible page improved or became a better match;
• search demand changed or became seasonal;
• the type of result users see changed;
• the query or its dominant interpretation shifted;
• your URL’s indexing, canonical or serving state changed;
• a confirmed search-system event coincided with the movement;
• the apparent change is partly or entirely measurement noise.

None of these explanations should be accepted simply because it sounds plausible. Each needs evidence.

First, confirm that the movement is real

Before investigating causes, check whether you are looking at a genuine and commercially meaningful change.

Compare more than one source and more than one date range. A third-party rank tracker may report a specific position for a selected location and device. Google Search Console reports impressions, clicks, click-through rate and average position across the data it records. These are useful observations, but they are not identical measurements.

Google explains how Search Console performance data is calculated, including differences in attribution and reporting. Its documentation also describes limitations such as anonymised queries, row limits, canonical-URL attribution and processing delays. Search Console’s data-anomalies guidance is useful when a recent or isolated observation does not fit the wider pattern. These limitations do not make a reported change false. They mean that isolated observations require verification.

Check the following separately:

• Is the movement visible across several days, or only in one collection?
• Does it appear in Search Console impressions or clicks as well as the rank tracker?
• Is the change present on the same device, country and search market?
• Has the query mix changed?
• Has average position moved because more lower-ranking queries entered the dataset?
• Has the business outcome changed, such as leads, sales or qualified visits?

Average position is an aggregate. It can move because the mix of queries, pages or search appearances changes, even when individual rankings have not all moved in the same direction. It should not be the sole basis for diagnosing a site problem.

This is also why ranking movement and traffic movement should be kept separate. A ranking can fall while impressions and clicks remain stable. Conversely, clicks can fall while rankings remain broadly steady if demand, query mix or the SERP changes. Our guides on rankings holding while clicks fall and content decay versus demand decay explore those distinctions in more detail.

Use timing and scale to narrow the diagnosis

Once the movement appears real, map when it started and how widely it spread. Timing is not proof of causation, but it helps separate possible explanations.

There is a meaningful difference between:

• one keyword moving: potentially normal volatility, a local SERP change or measurement variation;
• a group of related queries moving: potentially a topic, intent, competitor or SERP pattern;
• one page cluster moving: potentially a template, section, internal-linking or indexing issue;
• several templates or the whole site moving: potentially a broader technical, market or search-system change;
• rankings moving without impressions or clicks moving: potentially an aggregate or low-impact position change;
• impressions or clicks moving without rankings moving: potentially demand, query mix, SERP composition or attribution.

Break the data into query groups, page groups, countries, devices, templates and search appearances. Then choose comparison groups that are genuinely comparable. Unaffected queries or pages can act as controls, but only if they have a similar level of demand, intent, market and template behaviour.

A stable brand-query group, for example, may tell you little about a fall in non-brand category queries. A stable desktop group may not reassure you about a mobile-only change. Comparison groups reduce overinterpretation; they do not prove causation.

Test the main explanations against the evidence

1. Competitors improved

Your page can fall because another page moved up. This is a straightforward consequence of comparative ranking, although it does not prove that the competitor itself improved.

Look at the current results for the affected query group. Did one or more competitors appear above you? Did they add information, improve their category structure, strengthen internal links, gain relevant links or launch a page that appears to match the query more closely? You may not be able to prove which change mattered, but you can establish whether the competitive set changed.

Evidence that supports it: your page group declined while several competitors rose for the same queries; the SERP remained broadly similar; demand and your URL’s indexing signals are stable.

What could look the same: a search-result layout change may push competitors into more visible positions without a meaningful improvement to their pages. A different query mix may also make their apparent rise look larger than it is.

Action justified: inspect the competing pages and identify a specific gap or better match. Do not copy a competitor simply because it is above you. An unchanged page does not have a permanent entitlement to its old position.

2. Search demand changed

Lower demand can reduce impressions and clicks even when rankings remain stable. Seasonality, news, economic conditions, product cycles and changes in how people search may affect the available opportunity.

Compare the affected period with the previous period and, where possible, the same period in the previous year. Use Search Console impressions alongside external demand indicators such as Google Trends. These indicators are directional rather than proof of what happened to your particular audience. If impressions fall across several sites or the wider topic while rankings are broadly stable, demand becomes a stronger explanation.

Evidence that supports it: impressions decline without a corresponding position decline; similar pages or competitors show a comparable pattern; demand indicators or year-on-year data also fall.

What could look the same: a loss of impressions can follow a change in query mix, SERP features or Search Console attribution. External keyword tools and trend data are estimates or aggregated indicators, not proof of the cause of a specific site’s traffic change.

Action justified: avoid rewriting a page solely to compensate for a market-wide demand decline. Reforecast the opportunity, check the commercial impact and monitor whether the pattern persists.

3. The query or SERP changed

Search intent is not a hidden explanation that needs to be guessed from an algorithm update. It can often be examined through the results themselves.

Compare the current SERP with earlier records, screenshots or rank-tracking captures where available. Has the mix changed from category pages to product pages, from guides to videos, from local businesses to national publishers or from blue links to prominent result features?

A page can retain a similar numerical rank while receiving a different click opportunity. The same position means something different when the results above, below or around it have changed.

Evidence that supports it: the dominant page types or result features have changed across the affected query group; impressions or clicks change more than rankings; the current results answer a noticeably different version of the query.

What could look the same: a competitor may have improved, or the apparent pattern may come from a different device, location or collection method. Current SERP evidence may not reconstruct what users saw at the exact time of the original movement.

Action justified: adjust the page only if the current result pattern shows a clear, relevant mismatch. For a fuller treatment of this diagnostic, see our guide to detecting search-intent drift.

4. Your URL’s indexing or serving state changed

A page can be indexed without appearing for every relevant query. Indexing and serving are separate stages: being stored in Google’s index does not guarantee that a URL will be selected for a particular result set.

Indexing should therefore be investigated when the pattern points to a URL or page group, not used as a general explanation for every ranking movement.

Check relevant URLs in URL Inspection and review Page Indexing data. Look for changes in response status, canonical information, crawlability, rendering, robots directives, internal links and, where available, server logs. These signals can identify a technical issue to investigate; their presence alone does not prove that they caused the ranking decline.

Evidence that supports it: affected URLs show a changed canonical, exclusion, error, response or crawl pattern; the movement is concentrated in one template or directory; unaffected comparable pages do not show the same decline.

What could look the same: a whole topic may have lost demand or changed SERP treatment. A page can remain indexed and still lose visibility because another result is preferred.

Action justified: fix a technical issue when the evidence identifies one, then validate the affected URLs and monitor the page group. Do not make broad indexing changes because a rank tracker reported a lower position.

5. A confirmed search-system event coincided with the change

Google’s official Search Status Dashboard and related update records can provide useful timing and context. If a broad movement begins during a confirmed Google search update, record that relationship and compare your pattern with other affected query and page groups.

That overlap is not proof that the update caused your decline. Several causes can occur at the same time, and an update record does not reveal the causal response of every site, query or URL. Avoid naming unsupported ranking factors or treating third-party volatility commentary as a diagnosis. Check the official record again at publication because update histories and dashboard information can change.

Evidence that supports it: a broad, persistent pattern begins within the confirmed update window; multiple unrelated sections or templates move; there is no clear competitor, demand, SERP or indexing explanation.

What could look the same: seasonal demand, a major competitor launch, a reporting change or several local issues occurring together.

Action justified: use the event as context while continuing the investigation. Review affected and unaffected groups, search appearances and business outcomes. Do not publish a list of speculative “update factors” as a substitute for evidence.

6. The measurement is noisy

Search results are time-sensitive. They can vary by location, language, device, query context and collection method. A single low-volume keyword can move without a recoverable single cause.

Search Console also does not provide a complete census of every query. Anonymised queries, row limits, processing delays and canonical attribution can make relationships between queries, pages and outcomes less visible.

Evidence that supports it: the movement is isolated, short-lived, tool-specific or commercially immaterial; comparable data sources do not agree; there is no coherent page-group, competitor, SERP, demand or indexing pattern.

What could look the same: a real but narrow query change that has not yet appeared in aggregate reporting.

Action justified: extend the observation window, improve the comparison and check the business impact. Monitoring is often the most disciplined action, not a failure to act.

A synthetic example: one drop, several possible causes

Imagine an ecommerce site’s category page for “women’s waterproof hiking boots” falls from position 3 to position 8. The page has not changed.

That identical movement could mean several different things:

• Competitor improvement: two specialist retailers move above the page, while the SERP format and demand remain stable.
• Lower demand: impressions fall across the topic, rankings are mostly stable elsewhere and the same seasonal pattern appears year on year.
• SERP change: shopping results, review content or other features take more space, reducing clicks despite a modest position change.
• Indexing issue: the category URL develops a canonical or crawl problem, while related pages using the same template also decline.
• Measurement noise: the change appears in one rank tracker and one collection, but Search Console, other locations and later observations do not support it.

The number “8” does not distinguish these explanations. Timing, query groups, page groups, competitors, demand, SERP captures and URL-level evidence do.

This is a synthetic teaching example, not Liquid Silver client evidence or an estimate of how often each cause occurs.

Decide whether to monitor, investigate or change

After testing the explanations, make the decision explicit.

• Monitor when the movement is isolated, short-lived, tool-specific, low-impact or unsupported by a coherent evidence pattern.
• Investigate when a page group, template, market or query cluster shows persistent movement, especially if impressions, clicks or conversions are materially affected.
• Implement a change only when the evidence identifies a specific issue and the proposed change directly addresses it. Define how you will validate the result before releasing it.

There is no universal threshold for when a business should edit a page. A position change on a high-value commercial query may deserve attention sooner than a larger movement on a low-volume term. Consider persistence, demand, revenue or lead value, data quality, implementation risk and the cost of waiting.

When the pattern spans several templates, markets or teams, specialist analysis can be valuable because the answer may sit across Search Console, crawling, logs, SERP data, content, merchandising and commercial reporting. The benefit is not a more dramatic theory. It is a cleaner diagnosis and a safer implementation plan.

Ranking movement is information, not automatic proof of failure

An unchanged website can lose visibility because the competitive environment, demand, result page, indexing state, search context or measurement changed. That does not make every decline harmless. It does mean the first job is to understand what moved and why the evidence points in that direction.

Confirm the change. Map its timing and scope. Compare affected groups with credible controls. Inspect competitors and the live SERP. Check demand and URL-level indexing signals. Treat confirmed search-system events as context, not proof. Then choose the smallest justified action.

Liquid Silver can help when that diagnosis crosses site sections, markets or teams: connecting ranking, Search Console, technical, SERP and commercial evidence so the business can prioritise the issues that genuinely deserve implementation.