Qlik Sense and data preparation

Qlik has just released Qlik Sense 3.0. A major part of this release is the product’s support for self-service data preparation and it does one thing in particular that I have seen from no other vendor in this space. And this is with respect to data blending.

In most tools, when you want to blend two tables (or their equivalent if these are non-relational) from different sources together, you get one of two approaches. Some tools make suggestions – typically presented via a drop-down list – as to how the two tables might be joined. Some other tools don’t have this capability yet but plan to add it, while there are a few vendors (for reasons that I do not clearly understand) who think it is a bad idea and don’t offer it. In any case, how to blend data is either manual or you get a suggestion. Qlik also provides suggestions but instead of selecting from a drop-down list the approach is completely visual.

The best way I can describe this visualisation is to say that it is sort of like graph analytics except that instead of nodes you have bubbles, potential joins are highlighted using traffic light colours, and the bubbles float about your canvas and re-align to other bubbles as you make different choices (which you can back out of at will) and add more data to be blended. This description doesn’t do it justice: you really have to see it. It really is, seriously, impressive and a whole lot more intuitive than anything else I have seen in this space.

Of course the reason that Qlik can do this and nobody else can, is because of the associative database that underpins all the Qlik products. As a recap an associative database is a graph database that supports relationships (associations) between relationships as well as between nodes. This is actually much more efficient (you require far fewer nodes) than using a graph database per se, but it is also more complex, which is why associative databases have never really caught on for general-purpose use. However, that’s hidden from users in an environment such as Qlik’s, because the complexity is hidden from the user. In any case, the point is that Qlik can infer relationships between different datasets in ways that other tools cannot, and that’s why it is able to offer this graphical blending.

More generally there isn’t anything particularly outstanding from a data preparation perspective. It does have semantic capabilities built-in (indirectly reusing some of the semantic capabilities acquired from the company’s acquisition of Expressor) which puts it ahead of some but not all of its competitors and, of course, it has the advantage that it is built directly into the analytic and visualisation environment. However, the stand-out is the visual blending: you really have to see that.