How is "Previous Period" defined in Webmaster Tools?

[Jane_Doe]

If you enter in the date range to look at of last week in Google webmaster tools, what is the "previous period"? If you enter in the last two days in Google webmaster tools, what is the "previous period"?

For me anything from a week or so ago is showing a drop. Lately everything has been showing all green and moving up, but the reality is my traffic hasn't changed much at all these past few weeks.

I guess green is better than red, but I'm not seeing much improvement in my traffic overall.

[Planet13]

No idea... kept wondering about that myself...

[tedster]

By default, we'll show the last 30 days of data, along with how the daily average for the selected period compares with the daily average for the previous period.

[google.com...]
That doesn't tell us exactly what happens after we change the reporting period to something other than the default. I've always assumed that a period of N days, the change was figured over the preceding N days - but I can't find any documentation for that.

[aakk9999]

I have done some tests, filtering on only one keyphrase, that showed as only one query (i.e. the keyphrase chosen was not also part of another longer keyphrase). I followed 8 successive periods of 3 days (e.g. 20, 21, 22 Feb versus 23, 24, 25 Feb, versus 26, 27, 28 Feb etc). I have also chosen keyphrase that is not searched often, i.e. where numbers of impressions shown are relatively small (but still greater than 10 for any given day).

What I have found is that the comparison is done with the immediately previous period of the same length, based on impressions shown for this keyphrase below the chart. Also, the average position change is also done based on +/- from the average position in the previous period of the same length.

However I also noticed that:

a) If two periods have the same number of impressions, this is shown as "infinity" increase (where in reality should show as zero %)

b)Total number of impressions shown against that keyphrase was not always the same to the total number of impressions when I added up impressions for this keyphrase for each individual day within the chosen period that shows when hovering over the chart.

The rounding is strange as it does not follow consistent pattern, e.g. sometimes it appears that the impressions for each day within the period are firstly added up and then rounded, whereas in other periods it seems that each day is rounded separately and then daily rounded figures added up.

Regardless on how I split the periods, I could not get the consistent rounding formula used - it always failed on some period examples.

Overall over the 1 month shown on the chart, for all keywords (i.e. default chart that is shown), for this particular site, adding daily impressions showed approximately 30% the smaller number than the total number of impressions shown in bold underneath the chart.

[Jane_Doe]

Thanks for the detailed reply aakk999. So if I do the last week you think it is comparing to the week before?

And if I do March 3rd and 4th it is comparing them to the 1st and second?

I thought that might be what they meant but when I look at my numbers the don't always make sense to what my actual traffic and referrals have been. Webmaster tools will often show big ranking drops or increases across the board when the reality is my traffic has been pretty stable, except for day of the week traffic patterns, for 3 weeks straight.

[aakk9999]

Thanks for the detailed reply aakk999. So if I do the last week you think it is comparing to the week before?

And if I do March 3rd and 4th it is comparing them to the 1st and second?

This is what my testing showed, yes.

but when I look at my numbers the don't always make sense to what my actual traffic and referrals have been. Webmaster tools will often show big ranking drops or increases across the board when the reality is my traffic has been pretty stable, except for day of the week traffic patterns, for 3 weeks straight.

This is a different issue all together. My impression is that WMT is under-reporting the numbers - obviously, cannot comment on impressions or average position, but certainly it seems that when I add up click through . I have one micro site (6 pages only) with Statcounter which does not have lots of traffic so it is easier to compare the numbers - in fact it gets so little traffic that I can manually count clicks from Google reported in Statcounter versus report in WMT. From what I could see, WMT under-reports by at least 20%.

I was suspicious of this for some time, because I saw navigational queries with home page URL entered, that was reported to have only 20% CTR - which for navigational query seems far too low, considering the home page comes as #1 + 3 other results from the same site underneath. Even if a number of people may click on some of other result (such as alexa report etc), I would still consider CTR too low for a navigational query where the full URL is entered in google search box.