Best Business Intelligence (BI) Tools, Platforms and Vendors

Looker, Domo, Sisense, Tableau, and Power BI are tools are good for enterprise. If you are looking for SMB (Small Medium Business), where you have 5-10 users who use BI for portfolio, then you should look at small Startup (ConverSight.ai, ToughtSpot etc.) for a per-user license, which is comparatively low in the market.

This is a common question for many businesses that have gone down the SaaS route. SaaS Applications provide a critical application in the cloud based on a subscription model, which is great for companies looking to add software or shift capital expenses to operational expenses. However, the problem is that most of these applications are getting your data out of the SaaS and into a reporting data warehouse or another third party application.

The most common means of accessing your data is normally via an ODBC/OLE connection that is purchased at an additional monthly cost from the SaaS. This is normally a read-only connection but does provide a limited degree of access to your data. Don’t expect the SaaS Company to provide you with ER diagrams or table structures, nor should you assume the data model or schemas will follow any of the best practices or common design methods. They won’t. You must remember that SaaS-based applications try to fit everyone’s needs into a standardized set of tables. So expect odd columns and table joins throughout the application.

Once you figure out where your data is, you need to decide if you're just going to query the data from the SaaS database when the reports are run or if you're going to extract your data into a data warehouse nightly and then run your reports. This also means you now have to have a database server either on-premise or in the cloud.

I prefer the latter of the two options however it does introduce a gap in the data since the extracted data won’t be real-time. The benefit is that you can query the data as often as you want without impacting the live application. The most common tools normally used for this are MS SSIS and SAP Data Services. Both applications perform similar functions however there are many other ETL tools on the market.

Once you have the data extracted, your reporting software can run it reports. Gartner provides in-depth studies on the various applications however it really comes down to the following tools:
Microsoft Power BI
SAP Business Intelligence
Qlikview/Qliksense

These are the top reporting applications. Each has their pros and cons when compared however each can do basically the same thing as the others.

Here’s what makes them stand out from each other:

Microsoft Power Bi - Easy for end-users to create Adhoc reports from almost any dataset out there.
SAP Business Intelligence- Global standard for reporting, established tools and community
Qlikview/Qliksense- Top-rated dashboard and visualization tool.

Oracle EPM is more robust and provides comprehensive coverage for calculating KPIs. The products follow a global standardized pattern which has ease of adoption and implementation.

Power BI, Tableau, and Qlik have the capability to do calculations for KPIs, are easy to create visualizations, and also have the capability to get the source from any data sources. Qlik and Tableau have more advantage than Power BI. Qlik and Power BI provide the desktop for personal use. Tableau only provides trials. For the Enterprise version, it has a feature to share and collaborate with the team. Now, only Tableau can deploy the tools on Linux environments (Qlik and Power BI only for windows).

TIBCO Jaspersoft

I think Qlik is strong at data processing and has a high performance within a large scale of data with its unique in-memory database.

I have been doing data science by Python, now looking for a tools that can easily scale up the projects at low cost. KNIME seems to be a good one so I had tried it for a few months, but find difficult to copy my concepts in Python to KNIME.

Take 07_Customer_prediction_with_H2O as an example, I can code it in Python and well understand the concepts and codes, but when it comes to "nodes" in KNIME, I can't easily understand the relationship between nodes. I copied the sample process and modify a bit according to my data and understand the propose and concept, but I still don't understand why "Parameter Optimization Loop Start --> H2O Cross Validation Loop Start --> ......" in the Random Forest Metanode for instance, now I just copy to make it work but don't understand what reasons to put the nodes in that way.

Having said that I like the concept to make data science in nodes, as I am quite experience in playing around with Node-Red.

Domo or Pentaho and the differences between the startup version and enterprise version are the tools that you can use to pull out or get in the data and it is important to have the flexibility to get the information in different ways.

Total cost of ownership is often overlooked during BI product selection. Cheap products do not equal cheap ownership experiences, whether it is missing functionality which must be provided by additional products, poor integration of modules which causes duplication of effort, weak support from the vendor, high cost of maintenance or constant changes to the product portfolio.

The key factors to consider are:

Is this product right for the intended user-base? It should not be necessary to purchase one product for IT, one for business analysts and one for 'end-users'. There are considerable cost savings associated with using a single platform (not a single vendor with many products they have built or bought).

Does the product have the depth of functionality needed, and foreseeably anticipated?
Is that functionality accessible? Can an expert easily access complex, deep functionality, without the occasional or new user being overwhelmed by the interface?

Can it reach all the necessary data sources? Both inside and outside the corporation.

How fast is the user experience, both in developing reports and dashboards and in retrieving the data? Speed of both allows iterative learning and development by new and occasional users, while ensuring high productivity for expert users.

Does it work with our real world data? Too often evaluation of products still relies on superficial test on restricted volumes of data, or the lower complexity data as "it would take too long to build a fully representative testing environment" - big mistake. Identify the product(s) you believe are suitable and then bear the cost of proving they can deliver in your own use case. Too often a poor acquisition is followed by increasing spend to "make it work", when money spent earlier on selecting and proving the right tool would lead to much lower overall cost of ownership and more importantly early success and hence ROI.

Does the company have a history of good backward compatibility? You will build a vast amount of intellectual property with a BI tool. You will become dependent on the insight it provides your organization. So investigate how well you chosen product has allowed users to migrate that IP forward through new revisions of their products. Rewriting IP is a good opportunity to clean it up and start over, but it's a massive unnecessary expense if you have built what you need and it is the vendor forcing you to rewrite your work.

It is really difficult to choose the right one, in my experience, the most important thing is not the tool that we choose. The most important thing is to design the correct scope, what we need to give us the tool based on two factors, the data we want to obtain and for whom that data will be. If a good approach is made, the chosen one will be the correct one.

Use case! What and or for whom are you trying to provide reporting capabilities to and from what applications and data sources. What are the skill levels of these users - basic business users or analysts. Ease of use should be top of the list as well as a quality customer support experience.

There are many choices today around BI solutions (embedded, self-service, etc.) and finding the right one depends upon answering these questions.

Easy to use, performs well, cost, and not going to send me insane integrating it with other systems. Everything else is gravy.

Key Evaluation Points for BI Tools:
1) Ease of use – Self-service analytics and data exploration, End-user report consumption and Time to develop new reports
2) User Interface and Visualization effectiveness – If it does not look good business will not use it
3) Styles of BI Supported – Scorecards Dashboards, Operational Reports, OLAP Reporting, Predictive Analysis, Notification and Alerting
4) License/TCO – Open Source vs Enterprise Software – support and maintenance structure – development and administration costs, environment deployment constraints, admin tools
5) Performance and data scalability – Depends on the business requirements and data volume
6) Data source connectivity – Depends on the source systems that need to be integrated and backend data integration
7) Security – Authentication, Authorization, Application, Object and Data level security
8) Extensibility and API/SDK – If integration with external applications is required
9) Feature Set/Product Roadmap – scope of product features, speed of platform innovation

Most folks tend to think of UI and the flashy, sexy stuff. However, experience has proven that adopting a platform approach is key to success - a Platform that can access any data source and provide capabilities for internal and external users from the same platform, and not merely serve as a data visualization tool. The ability to do disconnected analytics cannot be discounted; Further having a platform that does NOT mandate creating a data warehouse as a prerequisite; i.e. It shoild have the ability to access operational data sources with minimal impact to the back-end systems.

I believe you have to match product capabilities to company BI strategy first and then consider infrastructure standards and internal competencies. if you do these things you should get value. The type of BI solution is important but don't be fooled by the "it's all about data visualization or big data". it is not in most organizations...

"Time to insight" is very important for me. Since the business questions arise non stop and need fast responses, going from raw data to answer a business question in the lesser time possible, with the easiest way of doing that, could be a huge differential in selecting a tool (set of tools).

That is no right answer for this. It actually depends on the needs, user capability, available infra and tools, and management's requirement. You can use Pentaho, Power BI or Superset if you are looking for free or cheap solutions. Else, you can go with Alteryx and Tableau, Enterprise Power BI, Qlik Sense, etc.

It all depends on couple of points, example: Targeted Users, Tool Usage, Budget for the tool,infrastructure etc.. Based on these factors you can decide about the tool. More or less all tools provide similar set of features. If you wanted to stay in Oracle area, you may use Oracle BI. Personally I like Power BI /Qlik Sense. Power BI is a very powerful tool for data analysis. However, along with Attunity and integration with snowflake makes Qlik sense a very good tool to be used on cloud.Also the insights from Qlik makes this tool great.

Currently, there are a number of different BI tools in a market that allows you to work with different data sources. In the Oracle world, there is an Oracle BI itself that is rather sophisticated and not as intuitive. Cognos requires a pretty complicated modelling and programming in order to use all its flexibility together with a limited performance when many big dimensions used together and a security that I had to rework and automate in order to avoid a time-costly administration (not talking about other inconveniences), some other products are not cost-effective and not self-serve.

I would prefer (and now switched to) a Microsoft Power BI that has a wide range of sources to connect to (if no driver issues), cost-effective, simple to use, highly graphical, cloud-ready, allows a self-serve in most cases, nice performance, continuously improving software and providing new capabilities and objects, simple security system, though based on a cloud may have some refresh delays. I would stay now with MS Power BI.

Restricting this to the Oracle products world, the standard choice would be OBIEE or the Oracle Data Visualization tool. Otherwise, there are plenty of other tools: Tableau, Qlik, PowerBI.

There are now good BI tools in the Oracle world. Go for an independent BI supplier like QLIK tech.

Depend of your use case, we are currently in an evaluation process for embedded analytics, the short list is: Looker, YellowFin, Domo, Sisense

I would definitely recommend IBM Cognos. I have worked with many BI solutions that were based on IBM Cognos and Oracle.

If budget is an issue then PowerBI is a very good solution too. I have also worked with some open source big data solutions - for reporting and visualizations Apache Superset and for OLAP Apache Kylin.

Power BI..Cost Effective, Self Service BI, Highly Graphical, Cloud Ready, Operational Report, Data connectivity.