Choose two adjacent ranges of equal length: for example, from 1 to 7 and from 8 to 14 of the month. The periods from 1 to 7 and from 2 to 8 overlap by six days, so the difference between their totals does not show a change between two separate weeks.
Check that the statistics are for the same site and cover both ranges. The import date is the day the information was received, not the start of a new period of clicks. If there is no data for the earlier period, a numerical comparison is not yet available.
Choose a specific metric
What do you want to understand: a change in clicks, impressions, or the status of a specific page? Do not combine all metrics under the heading “the site grew.” Each metric should have a definition and a period it relates to.
Clicks and impressions require an event date range. For page statuses in search, the time of the relevant source information matters. Do not treat a single total number of pages as a monthly total of impressions.
The weekly SEO report guide can help you prepare a clear structure. Start with the site owner's question, and only then choose data that can actually answer it.
Record both period boundaries
The period is shown next to Webmaster statistics in the reChecker interface. Save both boundaries in full. The import date is a separate label: it indicates when the information was received, not when each click occurred.
When calculating manually, include all the days in the range. A sample range from the first through the seventh of the month contains seven calendar days, not six. The next range of the same length is from the eighth through the fourteenth.
Do not choose arbitrary dates if some data has not yet been received. Any source limitations should be stated before the comparison. Do not silently fill a missing day with zero just to make the table symmetrical.
Avoid overlapping days
Ranges from the first through the seventh and from the second through the eighth differ by only two days at the ends. Most of the events are shared. The difference between their totals does not describe the change from one full week to another.
For two separate adjacent periods, the first ends the day before the second begins. They have the same duration. In the current dashboard, the Webmaster comparison uses such adjacent full ranges when suitable data is available.
If the dashboard says that the previous full period is required, do not describe yesterday's update as a replacement for it. Wait for a suitable source, or conduct a separately labeled manual analysis of the available portion.
Check the selected pair in a table
Let us choose several pairs for a fictional project. This is an example of period selection; no results for a real site are shown here.
| First range | Second range | Assessment |
|---|---|---|
| 1–7 | 8–14 | Adjacent seven-day periods |
| 1–7 | 2–8 | Six-day overlap |
| 1–7 | 8–17 | Different durations |
| 1–7, no data | 8–14 | The first total is unknown |
Account for events and use the same set of pages
Equal duration does not eliminate differences in business events. Holidays, a seasonal service, or the launch of a new section can change the conditions of a period. Record such events in an external log, without automatically presenting them as the cause.
If you are comparing a specific segment, use the same selection rule for both periods. The set of pages for one service should not accidentally expand in the second period without an explanation. Otherwise, the scope changes along with the figures.
The article on analyzing an organic traffic drop can help you frame follow-up questions. First confirm the change and the data included, then investigate possible explanations using relevant sources.
Check the calculation and the zero baseline
In a sample, the total increased from one hundred to one hundred and twenty clicks: the absolute difference is twenty, and the relative difference is twenty percent. If the first value is zero, the usual relative-change formula, which divides by that value, does not produce a usable percentage.
A zero total must be confirmed by received data. An empty source cannot be treated as a zero baseline. If the value is unknown, simply state that the numerical comparison has not yet been made.
For the sample, the final statement could look like this: “From 1 to 7 of the month — 100 clicks; from 8 to 14 of the month — 120 clicks; difference: +20 clicks (+20%). Source: Yandex Webmaster data for the selected site, covering two adjacent seven-day periods. The page list and comparison conditions are the same.” In your report, add the month, year, and date the data was received.
The results of checks are available in the site monitoring dashboard.