Monday, December 9, 2013

Terms Of BI / Busienss Intelligence Implementation - BI Adventure

Greetings. Still on the theme of BI Adventure, previously i discussed about BI FAQ, this time I would like to discuss about something different but still related. Today's discussion is about 1 of the existing FAQ. It is about main requirements of BI implementation.

As I have written briefly, the main requirements of BI are solid and integrated database. Why they are important? Because BI require data as capital, not file or information.

What is file / archive?
File is a set of related record. File also mean archive in Indonesian.

According to KBBI =
  1. Written documents (letter, certificate, etc), verbal or pictorial (photo, film, etc) from past times, saved in written media (paper), electronic (cassette tape, video tape, cd, etc), usually issued by formal institution or agency, stored, and maintained in a special place for reference.
  2. Standardization, regulation, and preservation of archival material that is needed in order to be recognized and organized as the original without anyone who tampered with and altered.
According to Oxford Dictionaries =
  1. A folder or box for holding loose papers together and in order for easy reference.
  2. A collection of information about a particular person or thing.
  3. Computing a collection of data, programs, etc. stored in a computer’s memory or on a storage device under a single identifying name.

Conclusion : file / archive is a "place" that used for "save" collection of "data or information" about something for documentation or "reference".

What is data? What is information?
  • Data = Is a fact that its shape can be text, sound, images, numbers, symbols, and so on, or the information is kept back as a fact that which has not had a clear meaning.
  • Information =  Is data that has been processed and has meaning that can be used for a specific purpose or decision making.
  • What report and dashboard will you make?
  • What will be your analysis?
Next question, if there is no database :
  • How to process data into information without database?
Last, if there is no solid and integrated database :
  • How do you ensure the validity of the data?
All 4 questions above are compulsory question that I always ask earlier to my client and user. Based on that experience, i classify client / user level according to their answers. Here they are :

Okay, you've got the data, now : Where do you keep your data? Archive or written? Or system? Or data warehouse?

Level 1 - Basic
Oh, keep calm... We store most of our company data in 'excel system', but some of them written or hardcopy.

A very classic answer that always we heard of and created another serial question :
What is excel system? Is excel a system? How big it's size? Some of them is subjective, how many?

In my perspective, microsoft excel is not a system or database. It is a product that created by Microsoft to help people process their data mathematically.

The collection of excel files can be tens, hundreds even thousands of files. It could also be in units of Byte, KB, MB, GB or even TB, etc. My experience when dealing with hardcopy or printed data, the number is not too much : can be dozens to tens for master data such as suppliers or customers, but these things can more difficult when it has several versions that we don't know whichever most valid version. It is called messy, while in IT language : not integrated. Hahaha...

So excel is better? Haha.. Who said that?

Imagine an excel file larger than 100MB. It is a torture for a less qualified notebook to open that file. If you can open it, certainly it takes time and you must close any other open programs. Hehe

Okay, back to our main topic : "data". Both excel and hardcopy, all contain data. One thing that I can assure you, that 9 of the 10 data in Excel are not integrated. The simple one is about naming : branch name, product name and business district code. All these things hard to be limited, especially it has no systems or manual processes.

Unintegrated data in excel file spawn another problem : how to unite all as one entity? Example : English Book or Eng Book. Which one is correct? We all can consider them as a single entity : a book, but system will consider it as 2 book. It was about master data. Can you imagine the transactions? Hahaha

If you still have all of these problem, it is wise if you do not implement BI before you optimize and create systems to support your business Application. SAP, Oracle, Microsoft Dyanmics or even Custom.NET are solutions that you can choose according to the budget you have.

Level 2 - Bloom
We have SAP for finance, AS400 for production and Acces for sales.

Well, looks better? Fine, we can skip several questions from level 1 and focus on the end : "Integration". Having many systems may not be an advantage if it is not integrated. Case : In SAP. product code use numeric (1, 2, 3) while in AS400 using letters (a, b, c) If 1=a, 2=b, 3=c, then it may be a great help, but is it always correct? In my past projects, everything was never the same and come in pairs. Hahaha.

For case like this, it needs extra effort to mapping or matching all master data in all running systems. How long? The answer is uncertain. Usually, each system is owned and used by multiple divisions. In this case, if it take 1 person per 1 system, we must hope that they understand all of the data. Another problems, rarely a person understand all at once, even though the System Support.

This kind of system is like a developing country, still needs a lot of repairs done internally. If the case like this, both users and consultants must cooperate to solve this problem, then proceed with the BI development. Once again without a strong and integrated data, the BI foundation will be weak and complicated, which eventually might become a waste.

Level 3 - Advance
There are several systems owned by our company : SAP, Custom .NET Applications and Access. For the current reporting we use  SAP as a reference for master data and transactions.

It appears that the BI project will not be too difficult. It is obvious that the the existing references is already clear. BI foundation already exists, so we can directly move to the next phase : to evaluate and test the feasibility for the creation of  a data warehouse. Do not forget to check back the data that also needed by the dashboard and reports later.

A few things to note is the level of integration. Integrated might be not complete, complete might be not integrated. Usually target or plan data have not been inserted into the system or database. Certainly, the business users will see their KPI achievement against their targets. Therefore it is required to make a simple application to store and stream the data into the data warehouse. At least, in this level, you will have a good start. Hehe...

Level 4 - Modern

Sir... For current reporting, we have a data warehouse retrieved from ERP System. The only matters, we feel that our reporting and analysis not really accurate, efficient and maximum.

I will be more happy when the client answer me with a sentence that has similar meaning with above sentence. Hahaha. The conclusion that can be generated from the user answer is : capital and container (database) are no longer an obstacle to the BI implementation and certainly clients need solutions to enable the business users to analyze and generate reports more accurate, fast and attractive. Please take a note that the existing data warehouse still needs to be evaluated. In addition to the BI product it self, the DW design and structure determine the performance and the speed of processing / query data.

Up to this time, I rarely find such a client. There is also a case that they already have other BI products, but the clients want a more powerful tools. Product benchmarking is needed to ensure that the solutions will be delivered by consultants align with the clients specification and requirements. Do not ever offer something that cannot meet the expectations or even worse from anything that has existed at this time.


In the end, I would just repeat that the foundation of Business Intelligence is a powerful and integrated database. They are absolutely something must be achieved before implementing BI product. Strong and integrated data will make the result more varied, attractive and inspiring, not just gossips that it's validity questionable. In addition, for the level no4 case, knowledge of BI product will be an unavoidable necessity. Last, I always say this everytime I post on this blog : All stories is based on my opinions and personal experiences. If there are shortcoming and mistakes, please forgive me and  I welcome you to provide any feedback on any post. Thank you.