Why Qlik is unique? Compare the data model with Tableau

People always compare Qlik (QlikView/QlikSense) and Tableau. From the functionality perspective, I would say Tableau and QlikView is more or less the same. However, Tableau has a better reputation on its code-free interface and sophisticated graphs, therefore it somehow lead the market as business people prefer a "simpler" solution.

When QlikSense first released in around 2014, I could imagine there will be another fierce competition in coming 5 years. At the moment QlikSense is at version 3, so it is not fair to compare with version 10 of Tableau. It still has a long way to go. Especially it is facing two huge obstacles - transform the current QlikView clients into QlikSense and Compete with Tableau.

But why I would say eventually QlikSense could be better than Tableau? The key reason is the data model behind the scene.

If you have ever used Tableau, you might be impressed by the user-friendly interface and the quick solution you are able to build. That's my first impression as well. But when I start to create a more complex analysis and the source data set grows, there are some problems appeared.

1. The data are actually not associated.

The blending mode is a virtual join of two data set. When you drag the first field to the tableau sheet, that source will become the primary data which is always on the left hand side. Then you drag another field from a different data source, that's the secondary data which you can imagine as a join with the primary data.

So when you do the filter or creating calculated fields, it leads to problems of data reference or duplication. The mistake is usually complicated to check or even recognize because the result is not necessarily to be stupid. And secondary data will cause some usage problem on Tableau (such as table calculation, filters, etc.). I saw there are some updates on version 10 specifically related to the data model part but at this moment I have not done any test.

2. You do not need a schema

Although star-schema may not be the best, it is easier to understand the model and the relationship between different tables. In Tableau, there is no such schema and you only need to aware of the primary and secondary data. This could be good because you do not need to plan in advance and just place the fields you need.

That comes with a problem. Fields could be duplicated and when the data set scale up, it can cost quite a lot extra effort to load those unused or duplicated data.

Compare to Qlik, Tableau requires a less structural solution which leads to a "simpler" software to most of the business people. As with Qlik, you will soon come across the loading issues such as "synthetic key" or "circular reference". Plus, you will need to use some codes to control the join or calculation during loading.

On the other hand, when you browse the dashboard, you will find a very interesting and useful feature that created from the "tough" model. The "Green, White, Grey" navigation method.

Green: Selected

White: Related

Grey: Unrelated

So the relations are easily distinguished.

There are some more features to differentiate these two products but I cannot go through one by one here (maybe in some future posts!). By my own experiences on both Qlik and Tableau so far, I believe Qlik has a stronger engine behind the scene but right now it's still the early stage of the product QlikSense. Tableau has more comprehensive features at the moment but still rooms of improvements on the performance and volume expansion.

Also, hope the other opponents such as Spotfire, Microstrategy or PowerBI...they can grow faster and bring more diversity to the market.

2.