A manager sees a page ranking higher on a phone than the report shows. A contractor checks the query in another city and gets a third result. Before discussing the quality of promotion, establish exactly what is being compared: the same query does not necessarily mean the same search conditions.
The basic unit of observation is a query together with the search engine, location, device, time, and method used to calculate the position. If any condition changes, it is a new measurement series. Keep the old history, but do not automatically call the difference between series a drop or an increase for the site.
Create a measurement-series specification
A single row in the specification should let another team member reproduce the check or understand its limitations. Fill in the fields before the first comparison, not after a disputed figure appears.
| Field | What to record | Why it matters |
|---|---|---|
| Search engine | Google or Yandex | Keep different search results separate |
| Query | The exact string, without editorial shortening | Prevent the wording from being changed |
| Location | The region checked and how it is set | Preserve the geographic conditions |
| Device | Desktop or mobile, according to the available settings | Avoid mixing result variations |
| Target | Domain and expected URL, if specified | Distinguish the site's position from the page's |
| Time | Date, time zone, and time of retrieval | Compare observations consistently |
| Method | Rank tracking, a manual snapshot, or an account | Avoid mixing different metrics |
Add the counting rule: what qualifies as a result, how additional blocks are handled, and how the domain's absence within the check is recorded. If the source does not explain this, mark it as unknown. A number without a definition is difficult to verify.
For a project covering several regions, create several specifications. Do not fill in the “region” field with the business's city if the query is actually checked from somewhere else. The project's intended area and the measurement conditions are different pieces of information.
Distinguish the site's region from the region being checked
A site's regional association describes how its offering is connected to a place. The region being checked describes the conditions under which the search results are obtained. Changing the latter does not mean the former has changed.
Yandex notes that regionality may be taken into account for geo-dependent queries. At the same time, association with one region does not rule out appearances in other regions if the content is relevant. This is a reason to compare local queries across regions, not a promise of the same position in every city. Yandex help on regionality.
In a hypothetical equipment repair project, the query “pump repair” is checked separately for two service cities. Do not try to fix a difference in results by changing the site's regionality at random. First check which addresses and service-call terms are published, whether the required regional page exists, and whether it matches the actual service.
In Google Search Console, you can filter data by country and device. The country in this report does not replace a city in a separate position check. Use the available dimension as stated; do not label it with a more precise location. Google documentation on filters and comparisons.
Do not mix a one-off result with average position
Rank tracking records an observation for specified conditions at a particular moment. A search platform account aggregates the history of impressions. A manual check adds another data point, whose conditions are sometimes only partly known. These sources can complement one another, but they are not required to show the same number.
Google explains how average position is calculated and how data is aggregated in the report. The metric depends on the set of impressions and the grouping method. It cannot be directly equated with a page's place in one manually opened set of results. Google help on Search Console data.
Keep these two questions separate: “How did the site appear for this control query?” and “How did its visibility change across actual impressions in the account?” Check the first using a snapshot or rank-tracking history, and the second using data from the relevant report with its filters preserved.
Do not replace absence within the check with an artificial position. If the tool does not find the domain, keep the status provided by the source. You can adopt a separate rule for calculating an overall metric, but describe it explicitly and apply it consistently to both periods.
Investigate a discrepancy by changing one condition at a time
When a position is disputed, save the original report and the manual result. Then compare the exact query, search engine, location, device, time, and URL found, one by one. Do not start by concluding that the tool has failed: first rule out obvious differences in conditions.
A simple pairing table is useful. In each pair, keep everything the same except one parameter: one query and region, but different devices; one device, but different regions; one configuration, but two different times. This shows what the observed discrepancy is associated with, although it does not prove the cause of the ranking.
Save the result URL. If a service page appears on desktop and an article appears on mobile, the same domain position can conceal different results. Note the expected page and investigate the URL change separately.
Methods and observation limitations are described in the guide to tracking website positions. To assess which elements occupy the results page, use the article on SERP analysis: an arithmetic position does not reveal the full context of a result.
Choose control conditions based on business needs
For each area of work, first select the actual service territories and the devices that matter to the project. Do not create numerous series just to make the report larger. Each series should support a decision the team can make if it changes.
For example, a local query is useful for monitoring the relevant offering and page, while a query without a city name may reflect a broader task. They can be monitored side by side, with separate rows for each. However, do not claim that one query represents all demand for the service.
Record the initial set and the rule for revising it. If you add a new region or replace the wording, note the date of the change. A new series does not inherit the history of a different configuration. Old records help explain changes to the report's scope.
Site Control in reChecker includes position monitoring in Yandex and Google. To compare history, preserve the conditions for the relevant check. A subscription lets you track changes, but does not mean that positions will automatically improve or be the same across different search engines.
Evaluate the setup using data, not matching figures
The setup is acceptable when the conditions are recorded correctly, the observation target is clear, and the result can be checked. An accidental match between a manual position and an account's average position does not confirm the quality of the monitoring as a whole.
Before acceptance, take several queries with different purposes. Make sure the history retains the search engine, configuration, and date; that the URL found belongs to the right project; and that missing results are marked separately. Check that the export or working spreadsheet does not combine different series under one label.
If the source does not provide some conditions, state the limitation in the report. A comparison is still useful within a stable method, but its scope is narrower. Otherwise, readers may treat as precise something that was actually measured at a more general level.
Set a response rule for changes
For a persistent discrepancy, check the specific page and whether it matches the user's task. For a one-off fluctuation, repeat the observation using the same configuration. If several series change at once, check the report's scope and broader site events before rewriting content.
In the final record, state the fact, the queries affected, the conditions checked, possible causes, and the next test. “The URL found changed in the mobile series for this region” is more useful than “SEO dropped”: it defines work that can be checked.
A specification and consistent comparison rules reduce disputes about figures. They do not explain the search engine's algorithm or guarantee a desired position. Their purpose is to make observations reproducible, so decisions about a page are based on comparable data.