High-performance, rapid analytics for corporate teams
October 28, 2019
High-performance, rapid analytics for corporate teams
Score 9 out of 10
Vetted Review
Verified User
Software Version
Qlik Sense Enterprise
Overall Satisfaction with Qlik Sense
We've used Qlik Sense Enterprise across a variety of departments within the organization as stand-alone installations rather than part of an enterprise program. These departmental implementations have typically been used to address business analytical needs such as market risk management, treasury operations, portfolio management, and inventory management.
It's been used as an analytical and reporting platform under full control of the business group/department to allow rapid development and iteration. The relatively small scale of the deployments (20-40 users each) has allowed this approach to work well and provide value. In addition to reporting use cases it has been useful for open-ended analysis.
It's been used as an analytical and reporting platform under full control of the business group/department to allow rapid development and iteration. The relatively small scale of the deployments (20-40 users each) has allowed this approach to work well and provide value. In addition to reporting use cases it has been useful for open-ended analysis.
- Qlik/Associative Engine - The data processing engine almost always performed extremely well. Many apps were around 50 million records with many fields with some over 100 million. Having this amount of data available to a moderate number of users for parallel, interactive/real-time analysis was quite powerful. Especially, because it was running on relatively inexpensive hardware without any effort expended on tuning or optimization.
- Built-In ETL - Having the option of connection to many data sources, and manipulating them with the load script proved to be extremely convenient and allowed quick development. There was a learning curve, and the script forced data processing to proceed in a particular way, but having the optionality was valuable. This allowed the development of end-to-end solutions without any back-end infrastructure and in many instances, none was ever needed.
- Fully Web-Based Development and Administration - Being able to do all development and administration via a web interface was a nice feature as it avoided having to install a desktop client, upload files, etc. Combined with the hub and the ability to copy app and sheets made the whole experience very fluid. In a very large scale deployment access to this would have to be managed carefully, but in our instances we could leave it open.
- Front-end control - In general Qlik Sense has provided less front-end/user interface control than the late versions of QlikView. Over time Qlik has increased the amount of control available to developers without having to resort to custom objects, but it has been a source of frustration. Every release does improve on this, but it still needs work.
- Information Density - In our environment, non-retail financial services, the typical need is for high-density reports/displays on desktop. This is one of the reasons Excel is so popular; it allows users to build high-density reports for themselves. It seems as though Qlik Sense was initially designed with tablets in mind, which resulted in a lower density than ideal. This in the process of being remedied with additional front-end control.
- Version Control - There was no easy way to version control data and application logic. This feature isn't something commonly available in this space, but would be very useful in running and controlling the platform. It's easy to generate many apps and data sets in the development and maintenance process and lose track.
- Analytics time reduction: 2 days to 30 minutes.
- In a particular instance we replaced an analytical process that required many iterations of data selection and model execution. Typically, this process will take 1 or 2 days consisting of correctly cleaning and selecting the data and manually running it through a model. It also required coordination between multiple people which contributed to its slowness.
- This process was rebuilt with a Qlik Sense app that was used for the initial data analysis/cleaning, the results of which were immediately sent to a Python model using the Server Side Extensions. The model output was then immediately displayed in the same app. This reduced the cycle to minutes and was run by a single person.
We primarily chose Qlik Sense based on our positive experience with QlikView. Qlik Sense seemed like a promising new direction based on a proven analytics engine.
Tableau was another option available to us but we didn't strongly consider it at the time since it didn't seem to have anything similar to the associative index, and the built-in in-memory database. Also, it seemed like it would really require a separate back-end (i.e. a database) to perform properly. It was definitely much stronger on the front-end/dashboard design side of things but we viewed that as something Qlik Sense could improve on over time. These differences are perhaps no longer significant now, but I cannot say as I have not done a recent comparison.
Tableau was another option available to us but we didn't strongly consider it at the time since it didn't seem to have anything similar to the associative index, and the built-in in-memory database. Also, it seemed like it would really require a separate back-end (i.e. a database) to perform properly. It was definitely much stronger on the front-end/dashboard design side of things but we viewed that as something Qlik Sense could improve on over time. These differences are perhaps no longer significant now, but I cannot say as I have not done a recent comparison.