Hi everyone,
Since the Web 2.0 boom, many of our users are companies that have websites which are the foundation of their business. These users often use Prism to analyze user activity on their site.
The table(s) they use to perform this analysis usually hold a unique identifier for a user (the email, for example), the company they belong to, etc. With Prism, It takes exactly 10 minutes to create a live dashboard that tells you, in real-time, exactly what you need to know about user activity.
However, a common problem these types of users usually encounter is that their table(s) contain a bunch of users that throw off their activity statistics - users created for QA purposes, fake users (users providing a fake email, for example), etc. This scenario has a very simple and effective solution. But before we get to it, I'll mention a few key features of Prism you should be aware of when tackling such a problem.
The Exclude Filter
The "Exclude" filter is very simple. It accepts a static list of members as well as results of other filters and returns all members that do not match these values. For example, if I create an Exclude filter over the Company dimension and specify "SiSense", this filter will return all company values that are not SiSense. We'll call this filter "Non-SiSense Companies".
Filtering By Label
Much like it sounds, the "By Label" filter returns members that match a text-based criteria. Creating a By Label filter over the Email dimension with the criteria "Doesn't End With sisense.com" will return all emails that do not end with "sisense.com". We'll call this filter "Non-SiSense emails".
Filtering By Attribute
The "By Attribute" filter is in a sense the exact opposite of the Exclude filter. In simple terms, it returns all members that intersect with the filter's defined criteria. For example, creating a Filter By Attribute over the Email dimension and specifying the "Non-SiSense Comanies" filter as its criteria will return all emails not associated with the company "SiSense". We'll call this filter "Non-SiSense Company emails".
Now back to the problem at hand...
Ok, so let's take a look at what we have so far. We have a filter that has only emails not associated with the company SiSense and another filter that has only emails that do not end with sisense.com.
You may ask yourself why I created these two filters that apparently return the same members. Well, they will not necessarily return the same members. It all depends on whether every person with a sisense.com email also entered SiSense as his or her company.
Regardless, whether we just pick one of these filters, or use an "Intersection" to combine them both into a single filter - let's call that final filter "Non-SiSense Users".
Prism Filters are just as valid anywhere a dimension is valid. This include on a pivot axes, in a formula, in a Drop-down List, etc. So instead of using the Email dimension in my dashboard, I will use the "Non-SiSense Users" filter. This will instantly cause my dashboard to reflect data relevant to non-SiSense users only!
Using filters in this way has two significant advantages, apart from focusing on real data and filtering out the garbage:
- This filters will always return up-to-date results. If tomorrow a new SiSense employee is hired, he will automatically be added to this filter (either by his company or his email).
- These filters can be modified at any time. I can modify the "Non-SiSense Companies" filter to include the company "SiSense Ltd" as well, without redoing any of my work and have it automatically update all filters, formulas and widgets that use it.
And to top it all off - any modification I make to these filters will automatically be reflected in the dashboard I created! I don't have to think of everything in advance, which at least for me is a very difficult task :-)
Elad
SiSense
For more information about filters, click here.
For more information about using filters in formulas, click here.
