Start by comparing equivalent data
Define the metric. An impression is a result being shown in search; a click is a visit through that result. An analytics system records activity on the website, while an enquiry or order is a separate target action. These measures are connected but do not have to match.
Compare periods with a similar meaning, such as several complete weeks with the same mix of weekdays. For a seasonal business, the corresponding period of the previous year can also help if conditions are comparable. A public holiday is not directly equivalent to a normal working day.
Record the dates of website updates, analytics changes, advertising campaigns and assortment changes. A different measurement setup can change the chart without an equivalent change in customer demand. Until measurement is understood, the effect of SEO cannot be assessed reliably.
Find the part of the journey that changed
Separate organic search from advertising, direct visits and other sources. Then compare Yandex with Google, mobile with desktop, and important regions and page groups. A decline in the overall total does not necessarily mean the whole website lost rankings.
Start with the largest absolute losses in clicks. A decline in a popular category may explain more than a large percentage change on a small article. For each substantial change, record the page, query or query group and the date the decline began.
This produces a short set of hypotheses: perhaps only mobile product pages changed after a theme update, searches for one product declined, or old URLs stopped receiving visits after a migration. These observations make the next check more focused.
Read impressions, clicks and rankings together
If impressions fell while comparable rankings remained broadly similar, investigate demand and the query mix. If impressions remained stable but clicks declined, review how the result is presented and how the results page changed. Neither chart pattern alone proves a cause.
Click-through rate, or CTR, is the share of impressions that produced a click. In an illustrative calculation, 200 clicks from 10,000 impressions equal 2%. One hundred clicks from the same 10,000 impressions equal 1%. This describes observed behaviour; it is not a universal quality score.
Average position also needs context. A different mix of queries can move the average even if some important terms did not worsen. Compare the same queries and pages, and do not judge the state of the business from a single average.
Compare losses with website changes
If the decline began after a release, inspect the affected page types. URLs, internal links, processing rules, main text or content rendering may have changed. For pages that disappeared from results, checking accessibility and URL status is a useful first step.
The history matters as well as the current response: what did the crawler receive when the change occurred? Yandex Webmaster provides crawl information, while Search Console offers inspection data and page reports. The team’s change log helps connect a technical investigation to a particular event.
Do not roll back the entire website solely because the dates coincide. First identify a change that could affect the lost URLs and assess dependent functions. Sometimes one template needs fixing; an unjustified complete rollback can bring back older problems.
Review demand and the usefulness of the offer
Seasonality, an ended promotion, availability and product interest can all change traffic. Compare query demand with what the page actually offers. A product that can no longer be ordered in the target region needs a separate decision about its range and content.
If technical barriers are not found, examine whether the page helps someone choose a service or product. Does it provide meaningful conditions, distinctions, current information and a next step? Which answers does the visitor still need to find elsewhere? This is more useful than repeating the target phrase more often.
Search systems and competing pages also change. A suspected connection with an algorithm update should be checked against official notices, but timing alone does not identify the required fix. Small fluctuations on a strong page do not justify immediately rebuilding everything.
Turn observations into a testable plan
A useful diagnosis ends with affected groups, verified observations, hypotheses and priorities. For each task, define the expected website behaviour and how to check it, such as a working redirect from an old URL or restored access to a category.
Set a separate monitoring plan after implementation. Rankings and visits need comparable queries and regions; enquiries need correctly configured tracking. A short period immediately after a change may not show a stable pattern.
If losses affect important services, several templates or a migration, a joint review by an SEO specialist and developer is helpful. General iT can start with an audit to identify the cause, sequence the improvements and define the scope instead of presenting an assumption as a finished answer.
Common questions
Does every traffic decline mean a penalty?
No. Check sources, measurement, demand, access and page changes first. If the owner’s tools show a specific violation, investigate it directly. The absence of a warning does not eliminate the need for ordinary technical and content analysis.
Why can traffic decline while rankings improve?
The query mix, demand or click-through rate may have changed. A site-wide average does not automatically describe each important query. Compare the same pages and phrases over equivalent periods.
Should we immediately rewrite every heading and page?
First identify the pages that lost traffic and the observations that are verified. Mass changes without a hypothesis make results harder to assess. Define a cause for each group and test a limited, relevant set of improvements.
Sources and documentation
From understanding to action
Connect the decisions in one plan.
A focused check helps identify a cause. When changes involve structure, development, content and analytics, agree them together with priorities, ownership and verification.
Tell us about your website, priorities and the issue you have observed. We will discuss whether you need a focused stage or coordinated promotion and what the initial scope should include.
Discuss coordinated promotion