Recently I was asked to create a Web Analytics dashboard on spec. I decided to use the latest version SAS Visual Analytics 8.1 so I could review the new features. When we wrote the Introduction to SAS Visual Analytics book, we were using beta versions of the application. Here’s some of the process I used while working on creating the Web Analytics report.
Getting the Right Data
The most obvious way to make your reports usable is to have the information your users want. For this Web Analytics dashboard, the organization provided their website data. They were not sure what to do with the information. For several days I researched what metrics to track! Then I had to determine if I had the data to support those metrics or could somehow generate it. You really do have to become an expert in your customer’s line of inquiry to help them. They may not always know the questions they want answered or they may be stating the question incorrectly.
Data from the web sites is mostly about a users visiting, how they got to the site, and what they looked at while on your site. Organizations want to make sure visitors get the needed information and determine how to better position their product or service. As smaller organizations work toward integrating analytics they often face the issue of not collecting the right data or having a way to capture the needed data.
This customer had been using Clicky to collect their site data. With the Clicky API, it’s easy to download the data into SAS! Since it’s an API – it will be easy to place this code into a batch process so the data is available every morning. Generally the team is only reviewing the data once a week so daily works fine.
There was a lot of raw data available so I spend several days changing URLs into meaningful analysis. For instance, the data contains the Referrer Domain – which is how the user was directed to the site. It’s useful to understand how your visitor arrives, but you want to have it grouped. I suspect Google is the main referrer but I can optimize the traffic for all Search Engine traffic if I understand it’s my main source of traffic.
I created a macro to scan the text and assign the traffic source.Later I’ll change this to use regular expressions but when I’m discovery process, it’s easier to do it this way. It allows me to move the “or index” statements around if I find a new traffic source.
Don’t Keep Your Draft Report!
If I see report builders do anything wrong – it’s always the same thing. They build a report without any idea of what they want to learn or what the report should say – then they keep that draft report! This is a mistake because the report most likely has artifacts from the development process that can cause issues later. Your first step is listing the questions you want to answer! Later you’ll want to break that list into areas of interest. For instance with web analysis, it would break into traffic analysis, visitor analysis, campaign analysis, and page analysis.
Proving Theories with the Data
When you have sample data, SAS Visual Analytics makes it easy to do some exploratory analysis.I wasn’t sure which of the fields would work together and which were not needed. My process was a little slow because I would change the data, reload to SAS Visual Analytics and work toward the next question on my list.
Along the way I discovered the data contained items that prompted some other questions.As stated earlier, I didn’t plan on keeping this draft report! It is easier to rebuild a correct report than muddle through my drafts to determine what data was needed and what was extra.
SAS Visual Analytics makes it easier to see where the data may be leading you down the wrong path. The customer was concerned that visitors were coming to the homepage and leaving immediately. It was easy to build a chart and see the root cause of the issue. It was true that site homepage received many visitors who left in about 15 seconds! Eeek!
These weren’t real visitors. Turns out the organization’s timesheet application redirected internal users to the homepage after they submitted their timesheet. These visitors would then click away to continue their activities. So I had to create a filter for those internal users.
Planning Your Report with the Data
With SAS Visual Analytics, I easily created custom categories. This helped me many times. The customer had run several campaigns. Here’s an example of where I used a Custom Category to group the Campaigns. The campaigns were a mixture of newsletters, social media, and planned events. The campaigns had to be based on campaign type! SAS Visual Analytics made it easy for me to categorize the values into larger groups.
This quick categorization helped me determined how to parse the field with code, but this method made it easier to determine the data hierarchy without writing a dozen queries in SAS Enterprise Guide. The speed of SAS Visual Analytics made it easier to do the work.
As I worked through my report, I created a table that listed what I expected each page to contain. Here’s an example of my layout planning. The data items column is key because it would help me plan what each source data set should look like. This way I ensure that I don’t have any unused data. Plus it’s a nice for your end user to have documentation about the report. From this information I will create the final datasets and final report. The draft report will be sent to the graveyard.
Reviewing the Draft Report
Here’s part of my draft report. The first page of this report is about traffic analysis. All of the charts describe when the site was visited, how did the visitors get there, and what they did while at the site. You can click on the line chart to filter the charts below. This gives the user an idea of what happened that particular month. This report most likely will be updated to only have the past 12 months when it goes into production. It’s really not relevant what happened on the website three years ago.
I haven’t decided if I’m going to keep the filter at the top – right now it allows you to see Customer, Internal User, and Job Seeker traffic. I suspect the report viewers only care about the customer traffic.
I’m a long way from finished with this report but SAS Visual Analytics is making it easy to explore the data and determine how to best setup this report for analysis.