In this post I will talk about what I think it is important about analytics tools. When I say analytics tools I’m not just talking about those tools that analyzes the behavior into a particular website (whatever you call a website) like Google Analytics, Yahoo! Web Analytics or Omniture.
I’m talking about all the Analytics tools, included the other ones that analyzes Behavioral information like campaigns, A/B testing and email marketing. Also the ones that analyzes attitudinal like Social Media Analytics and surveys.
The process begins with information retrieval. All the above mentioned tools has a particular way of doing that, even tools that do the same work, at least, slightly different. That difference would generates small or huge differences in the results, so it’s extremely important to understands that process.
The next step is processing the information. Normally the retrieved information is saved in a raw database, but that database would not be used for reports. The amount of information could be huge and even a simple query can take hours.
Can you imagine using an Analytics tool and when you try to get a report it takes 2 hours? mmmhhh, I don’t. Even when I remember my self using an Analytics solution (it was in 2004) that every time I dropped a Query, it generated a message like “your report will be done in 6 hours”….I don’t miss those times, not at all!
From the above mentioned raw database normally the information is processed and saved in a new database that can be queried, dropping results in an acceptable time frame.
What happened between the raw database and this second database (data warehouse)? The information is pre processed based on the limited reports that you will be able to get from the Analytics Tool. I mean, the raw database is more flexible. Remember some previous post talking about Flexibility vs. Easy to use? Well, the raw database is the most flexible way of analyzing information. You can basically analyze the information as you wanted to. But, it is not friendly and it is not fast. The data warehouse it is less flexible because you can’t analyze any information you wanted but just the available one. The available one is the previously processed one. If the information that you are looking for was not previously planned and processed, the information will not be available, at all. So, you can say that the data warehouse would be a brief of the “census” based on what the people that designed it thought it is important…so if you just ask about the profile of the people doing this you will get a very interesting clue about what can you expect about the solution.
So, now what? Well, we need reports. I’ll give you the recipe for a report which is at the end a data table that will or not be represented by a chart. You will need two ingredients:
1. Metrics (or indicators or KPI’s): “What we wanted to know”. Visits to a website, Mentions to a brand, emails opened, impressions, CTR (click through rate), etc.
2. Dimensions: How the metrics are splitted or distributed. It could be By day, by country, by language, by new or previous visitor, etc.
By the combination of metrics and variables an analyst can get or a very useful insight or just data. All the above mentioned tools have normally a set of standard reports, it is a pre defined combination of metrics and dimensions that their team identified as important or useful. Even when those pre defined reports can be somehow useful the possibility that your particular information need, in a particular time matches with what the team from the vendor company thought could be useful is nothing but utopian.
That is why the most advanced tools have something called Custom Reports. Custom reports, even when will not allows you to query the raw database, will allow you to combine all (or almost all, since sometimes some queries even in the data warehouse are pretty heavy) the metrics with all the dimensions. This way the possibilities of getting useful information for a particular need are higher.
So my suggestions are:
1. Never use a platform that doesn’t explain its methodology.
2. Never use a platform without Custom Reports. It can be somehow interesting, but you will use it for making decisions and not for fun.