Web Analytics TV #24 – The Ninja Show
March 16th, 2012 | Published in Google Analytics, Uncategorized
Welcome to the ninja show, aka Web Analytics TV! Web Analytics TV, as you well know by now, is powered by your amazing questions. In this awesome episode we had questions from Dubai, India, Germany, Sweden, France, Norway, Sweden, The Netherlands, and the US.
If you’re new to this show, our process is simple.
Step 1: You ask, or vote on, your favorite web analytics questions. Vote on next week’s questions using this Web Analytics TV Google Moderator site.
Step 2: From a secret undisclosed location at the Googleplex Avinash Kaushik & Nick Mihailovski answer them. :-)
In this episode we award the “Ninja of the Episode” and award it to Joe for a great question about the difference in data between the Visits to transactions report and the multi-channel funnels path length report. Joe, just email us and we’ll send you a signed copy of Web Analytics 2.0.
OK. Here is the list of last episodes questions.
In this action packed episode we discuss:
Here are the links to the topics we discuss:
In the fourth question we answer if sampled (fast access mode) data uses a random set of sessions. Just to recap, in most reports, Google Analytics pre-calculates report data and there is no sampling. But in some cases, like when you apply a secondary dimension, the report data needs to be calculated on the fly. In this case, Google Analytics might only process a subset of sessions so that it can return the report in a timely manner. Here’s how the subset of sessions is determined.
Currently fast-access model (sampling) can be controlled via a slider and the default is to process 250,000 sessions. This means that in the date range, only 250,000 sessions will be used in the report calculation. In Google Analytics, visits count sessions. So say a site received the following 3 days of visits.
Day 1 : 500,000
Day 2 : 250,000
Day 3 : 500,000
So, a total of 1,250,000 visits occurred in the date range. To determine which sessions are selected in the fast-access sample, Google Analytics first calculates a multiplier by dividing the total visits in the date range with the number of visits selected on the slider.
multiplier = 1,250,000 / 250,000 = 5
Then for each day, the total number of sessions is divided by the multiplier and only that number of sessions is randomly chosen from each day to be processed.
For Day 1 : 500,000 / 5 = 100,000. So 100,000 sessions are randomly chosen from day 1 to be processed to generate the report data for that day.
For Day 2: 250,000 / 5 = 50,000 sessions are used
For Day 3: 500,000 / 5 = 100,000 sessions are used
Once Google Analytics has determined which sessions to use, the sessions are all processed into a report. Then the report values are then multiplied by the multiplier to scale the final results.
Whats nice about the new fast-access mode slider feature, is that you can use the slider to change the number of sessions used in the processing. And by comparing the number of sessions used, you can see the impact of sampling on your data.
Ok great question.
As always, if you need help setting up Google Analytics or leveraging the advanced configuration options, we recommend hiring a Google Analytics Certified Partner.
If you found this post or video helpful, we'd love to hear your comments. Please share them via the comment form below.
This series would not be possible without your awesome questions. Please submit them on our public Google Moderator site, and while you’re there don’t forget to vote for your favorite questions. Avinash and I will answer them in a couple of weeks with yet another entertaining video.
Thanks!
If you’re new to this show, our process is simple.
Step 1: You ask, or vote on, your favorite web analytics questions. Vote on next week’s questions using this Web Analytics TV Google Moderator site.
Step 2: From a secret undisclosed location at the Googleplex Avinash Kaushik & Nick Mihailovski answer them. :-)
In this episode we award the “Ninja of the Episode” and award it to Joe for a great question about the difference in data between the Visits to transactions report and the multi-channel funnels path length report. Joe, just email us and we’ll send you a signed copy of Web Analytics 2.0.
OK. Here is the list of last episodes questions.
In this action packed episode we discuss:
- (0:28) Determining what caused your visitors to become return visitors
- (2:25) Calculating click though rate on site visits
- (3:43) Best way to report on internal user traffic to your site
- (5:23) Does sampling use a random set of sessions
- (6:20) Differences in visits to transactions and path length in multi-channel funnels
- (9:20) Using events in a goal funnel
- (9:46) Best ways to debug iOS and Android apps
- (10:37) Calculating number of events happen per session
- (11:40) Best practices for tracking duration in a video
- (12:43) Differences in how JavaScript and Android libraries persist visitor info
- (13:35) Differences in unique visitors and repeat visits
- (14:43) Impact of setSessionCampaignTimeout on multi-channel funnels
- (15:45) Importing conversion goals from AdWords into Google Analytics
- (16:48) Viewing Google Analytics on an i-device
- (17:53) Using outbound link tracking to track exit links
Here are the links to the topics we discuss:
- YouTube video demoing how to debug Google Analytics in Android applications
In the fourth question we answer if sampled (fast access mode) data uses a random set of sessions. Just to recap, in most reports, Google Analytics pre-calculates report data and there is no sampling. But in some cases, like when you apply a secondary dimension, the report data needs to be calculated on the fly. In this case, Google Analytics might only process a subset of sessions so that it can return the report in a timely manner. Here’s how the subset of sessions is determined.
Currently fast-access model (sampling) can be controlled via a slider and the default is to process 250,000 sessions. This means that in the date range, only 250,000 sessions will be used in the report calculation. In Google Analytics, visits count sessions. So say a site received the following 3 days of visits.
Day 1 : 500,000
Day 2 : 250,000
Day 3 : 500,000
So, a total of 1,250,000 visits occurred in the date range. To determine which sessions are selected in the fast-access sample, Google Analytics first calculates a multiplier by dividing the total visits in the date range with the number of visits selected on the slider.
multiplier = 1,250,000 / 250,000 = 5
Then for each day, the total number of sessions is divided by the multiplier and only that number of sessions is randomly chosen from each day to be processed.
For Day 1 : 500,000 / 5 = 100,000. So 100,000 sessions are randomly chosen from day 1 to be processed to generate the report data for that day.
For Day 2: 250,000 / 5 = 50,000 sessions are used
For Day 3: 500,000 / 5 = 100,000 sessions are used
Once Google Analytics has determined which sessions to use, the sessions are all processed into a report. Then the report values are then multiplied by the multiplier to scale the final results.
Whats nice about the new fast-access mode slider feature, is that you can use the slider to change the number of sessions used in the processing. And by comparing the number of sessions used, you can see the impact of sampling on your data.
Ok great question.
As always, if you need help setting up Google Analytics or leveraging the advanced configuration options, we recommend hiring a Google Analytics Certified Partner.
If you found this post or video helpful, we'd love to hear your comments. Please share them via the comment form below.
This series would not be possible without your awesome questions. Please submit them on our public Google Moderator site, and while you’re there don’t forget to vote for your favorite questions. Avinash and I will answer them in a couple of weeks with yet another entertaining video.
Thanks!