If the page is working and its URL has not changed, start with several measurements using the same query, city, and device. One bad day alone does not explain what needs to be fixed. Check immediately if the page is unavailable or has been accidentally blocked from indexing.
Suppose the rankings for one phrase across four checks were 12, 14, 11, and 13. That is a fluctuation within a small range. If a directory appeared instead of the service page, or the URL stopped opening, investigate the specific change to the page.
Record measurements and missing data
Record the exact phrase, search engine, region, device—computer—and selected group. For each measurement, save the date, the ranking if available, and the URL found. If any of these conditions changes, mark the start of a new series.
Do not combine data from different search engines into one series. Yandex and Google provide separate observations, and some queries on a schedule may be checked only in Yandex. A similar number in another system does not make up for a missing measurement in the first one.
For a missing result, leave the corresponding marker. Do not replace it with the previous ranking or an assumed last position. This substitution makes the series look smoother, but hides an important limitation in the data.
Choose a period based on query frequency
Daily, weekly, and monthly groups produce different numbers of data points. A short period may contain enough observations for a daily query and almost none for a monthly one. Your conclusion should account for the actual frequency, not just the selected date range.
The project supports up to fifty daily queries, up to one hundred and fifty additional weekly queries, and up to eight hundred additional monthly queries. Yandex follows this distribution; Google uses the daily group. A scheduled check does not mean that every check will finish at an exact time.
Check the date of the latest data point. If the series ends with an old snapshot, state that in the report. A recent date for viewing the dashboard does not update the timestamp of the underlying data. A new available measurement is needed before discussing the current state.
Record releases alongside measurements
| Measurement | Observation | Page event | Permissible conclusion |
|---|---|---|---|
| First measurement | Ranking found, URL known | Baseline state | An observation starting point is available |
| Next check | Ranking changed slightly | No edits made | A fluctuation was recorded |
| After release | The same URL was found | A technical defect was fixed | The work is accepted separately from the ranking |
| Later measurement | Ranking improved | Conditions remain the same | The ranking is higher with the same settings |
| After changing the region | Different data appeared | New setting | A separate series is needed |
If the team had several releases during the period, record all of them. Selecting only one convenient change creates a false explanation for the result. Also note changes to third-party components as a limitation if they cannot be reconstructed.
Check whether the URL changed
The URL for a single phrase may change. In that case, a domain ranking line hides a change in page content: visitors have started landing on a different section. Read this separately from a small change in the number and assess it against the person’s task.
Open both URLs and check their purpose, service terms, and next step. Do not delete material just because a different page than the expected service page appeared in search. The pages may need clearer roles, or their technical signals may need to be checked.
If the URL is missing from some data points, preserve that uncertainty. You cannot attribute the entire series to a page the team selected later. For an analysis of a specific service page, see the guide to analyzing search results.
Open an audit for the relevant page
A page audit has a different date and purpose from a search measurement. If you find an indexing block, record when you checked and the URL. After fixing it, confirm the technical behavior. Only then should you look separately at subsequent search snapshots.
The overall technical score for the entire report does not explain the movement of a specific phrase. Check whether the service page was included in the audit and which findings applied to it. If the page was not checked, state that instead of concluding that there are no errors.
To clarify the page’s state, use the technical audit. Report the result as a defect fixed or content improved. Keep the ranking as a separate observation, because search has other factors too.
Choose the next step based on the data
State what is confirmed: the conditions stayed the same, the same URL was found, a technical defect was fixed, or a particular direction is visible across several data points. Then list what remains unknown separately: incomplete data collection, a changed list, or a page that was not checked.
The next action should address this data. If the problem can be reproduced on the service page, assign a fix. If there is no technical basis for action, review how well the offering matches the query. If the settings have changed, start a new measurement series and record the query, city, and device.
For a longer-period overview, see the guide to a quarterly website review. It helps organize decisions, but does not replace the specific source data for the selected phrase.