Subscribe to Dr. Granville's Weekly Digest

Every BI product today is claiming to support self service BI. What does it mean indeed? Is it possible to map complex BI operations and data models to a self service BI user interface? Self service BI claims that IT support would not be required to run these applications. Can someone build complex SQL query without knowing SQL language? . To build the data required for a simple insights, analysts have to run lot of analysis on the data. How self service can help them out? May be it could be possible to map simple BI operations to some user interface for quick access on analysis but again it will require training from the technical staff. I am not sure how IT could be exempted from self service BI support. For example lets say we create a simple query builder drag drop interface to build the SQL query. Do you think an analyst who do not know simple SQL constructs can use that interface to build his data query.

Tableau claims to be the very first self service BI product. I am not sure if a non technical staff can use tableau to build an insight from any complex data. It seems that IT support is mandatory in any case. I have been using tableau for a while now and i had to build an analysis from multiple data sources by using their data merging technique. it was hard time even for me (a technical guy) to merge the data and to build an analysis on top of it. How self service can replace IT, its not understandable. 

To me, self service BI can only be possible if an application provide end to end easy to use user interface for non technical user. may be a manager would like to explore the data but he do not know SQL Queries. I am not sure if such a user interface is available  at the moment or it could be provided.

Do any big data analysis product supports such an intuitive user interface? . I will keep this question open for the community to answer.

Views: 825

Tags: BI, Hype, Self, Service


You need to be a member of Data Science Central to add comments!

Join Data Science Central

Comment by Lucky Balaraman on September 2, 2014 at 9:08pm

Muhammad, thanks for starting off this no-nonsense, insightful thread. It's rare to find such unbiased opinions on BI tools.

Michael Alexander, your comment exudes much common sense, especially where you say that the business user really doesn't care which visualization tool is used, as long as it gives him the insight he needs.

Comment by Homer Roderick on August 12, 2014 at 5:41am

Adding insight from the trenches to Peter's comment...IT's inability to keep up with timeframes and reluctance to embrace cloud options and unvetted external data is the driving force behind Self Service DI and will lead to a diminished IT role in BI in many companies. ITIL is just not appropriate for every endeavor.

I expect cloud options will emerge outside the corporate firewall to answer the hot questions and mature to become the goto source of knowledge. 

Comment by Peter Evans on August 12, 2014 at 4:25am

Hi Muhammad, thank you for your kind words.  I agree if we are talking about ETL processes that these should be hidden and be part of the routine that provides the data for the visual data story.  However one of the main reasons for the success of Visual Data Discovery tools being provisioned by the business in the last few years is that these tools appeal to the Business User in that the provide Self Service DI most of them call it Data Merge or Data Blend.  Business users are always looking for another way to slice and dice - and some teams are not even worried that the data is perfect when they do this.  This leads to shadow IT - it should be the IT department that leads this provision of Self Service DI enabling the user to add data into the DW/BI data that has being provisioned out to them - with the appropriate warnings on accuracy - this prevents the business user from becoming frustrated and provides a collaboration route.  This collaboration route can take many forms but usually drives data from outside IT to be retro included into the ETL process and being brought under the MDM umbrella if it proves to be of use to the business.  The Business understands process but also understands that to keep up to date and provide the best 360 degree view of their data that they need flexibility - ETL process deliverables cannot sometimes keep up with this timeframe - hence Self Service DI - which leads to collaboration and better Enterprise Data for all if implemented correctly and supported by the whole team IT and the Business.

Comment by Muhammad Javed on August 11, 2014 at 11:50pm

everyone, thanks for the precious thoughts! , in my perspective Self service BI and Self service DI are two different domains. Self service BI can be mapped to questions/answers and self service DI can be mapped to story telling. For example i should be able to build a visual data story and all the ETL routines should be hidden behind that story. A story could be "NY population that is complaining more about Iphone5". I should able to configure the data sources from where this story should fetch the data and also should be able to define the target of the story where it should store the data. think of everything as a story in business. ETL is something technical, why a self service should take care of ETL? its is responsibility of system designer /data scientists/architects to hide all this information in the background and expose whatever business needs.

Make sense?

Comment by Peter Evans on August 11, 2014 at 6:54am

Hi Muhammad, excellent points and right on the target.  I do also believe, like you, that if we are ever to have a successful state of 'Self Service BI' that 'Self Service Data Integration or DI' needs to be part of the mix. You talk about the end to end solution and this is what Self Service DI would allow.  In the past when end user reporting was the goal it was fine to deliver to a customer or client a tool that would allow them to take a prescribed set of data - created and delivered via either IT or the LOB Power User - and slice and dice that data to get results that they could interpret and use in their decision making.  However with business moving as fast as it does now it is imperative that the end user has the ability to join extra data into that delivered prescribed data set to change the focus.  This means that they need the ability to join data from the single source of the truth (or many such sources) and then add in data that has only just come into the operational arena and may not have being washed through the companies ETL process.  This is scary to some because it removes control from the Data Stewards however to prevent business users from going off and creating their own data lakes and ignoring the work IT has done it must be a requirement I believe in all organizations going forward.  With Analytical packages now becoming far easier to use and pre packaged it is up to us all to enable end users to use these methodologies in the easiest way and embrace the future whilst maintaining the security inherent in our current systems.


Comment by Homer Roderick on August 8, 2014 at 6:35am

Drawing from some of the futuristic thinking in the book "Average is Over" I suggest that when we can ask a system "How i can increase the profit in business?" and get an answer then we no longer need the person asking the question. Presently we still have to use human reasoning to answer questions like "With my limited resources and a defined set of options, how can I make the most profit". Machine assisted analysis (BI, DataScience, buzzword3.0) can help with that answer and in some cases answer it completely.  What we still depend on human reasoning to do is define the set of options. 

A tangent remark to Brad's comment, there's the data ETL that needs to be done. But who does the work (IT or ?) and where it is done (corp network or cloud or desktop) is a business decision. Especially when pulling in external data, but also with corp data, I've found working for the data consumer department in the cloud exponentially more productive than working within IT on the corp network.

Comment by Brad Sheridan on August 8, 2014 at 4:28am

Muhammad - great post that really calls out one of the most hyped parts of the industry right now.  IT involvement will always be needed, a lesson I learned 15 years ago when a vendor came in and sold some finance users and application that "doesn't require any IT support whatsoever".  Needless to say, I spent about 20% of my week supporting the application and the users.

We too are Tableau users and the way that I have enabled self-service for our users includes several steps: training at beginner, intermediate, and advanced levels for the business users on not only the tool, but the concepts of the visual communication of data as well, and by utilizing an agile project methodology which includes the business users early in the process which enables a level of familiarity and comfort.  These efforts, along with other enabling steps, allow my team to have more time to create the data models to support the business users.  We can package data up as an extract on the server for them to access, we can materialize the data in the database, etc... This all results in better governed data that the users can trust and make sound decisions with, all while obfuscating the complexity of multiple data sources.  For advanced analytics and basic data discovery (which usually indicates datasets that are random in nature), we work with the business analysts that have the right blend of technical aptitude to compliment their business domain knowledge (this is where the advanced training comes in)

Comment by Michael Alexander on August 8, 2014 at 2:35am

Homer, very well put. I don’t think I have heard such a solid answer in years as to the breakdown of Self Service BI.

Muhammad, you also have some very good points, and is one that has actually been addressed with the new SharePoint 2013. Such as Google, SharePoint also allows you to query data in a similar manner (ref. I would expect in future editions, we may see this expand so that touch screens/kiosks will also interact with data. As for a Data Scientist/Self Service BI individual, I can tell you that sometimes, we try to answer the questions that the business owner has, but many times, our line of communication is thru other departments who report the data in a way that makes sense (usually with the aid of an SME). These are the individuals we cater to, and whose lives we try to simplify. We realize that these are the individuals that are on the battle front, hit with question after question, one report always leading to another. Data is needed, we provide. In turn, this data raises more questions, we then provide more data. We try to find new ways to simplify the answers thru the use of an ever increasing BI stack, so that each department can make decisions more wisely with facts on data they are reporting against, providing quick results to the business owner(s), but alas, I do not think there will ever be an “all encompassing” report that will capture all the data needed for an Owner/EVP meeting. We however, can certainly try and provide as much as we can to aid each department so that they may contribute in the decision making process thru solid facts that are readily available.

Comment by Muhammad Javed on August 8, 2014 at 1:03am

Everybody, Thanks for sharing your thoughts. As for as true definition of BI is concerned, every business has standard set of problems of which people want to find answers of. The success of every business is measured from its profit. The first question every business man asks is "How i can increase the profit in business?" in my perspective all of the data science and BI revolves around to answer this question. Data Scientist try to find the correlations/patterns/anomalies in the data that can help to find the answer of this question and Analysts trying to analyze the problems in the same data. In my perspective self service BI is about providing the ease to business man to find the answer of the question he asked. Business man knows about his business and the data about his business, why he need data analysts and data scientists to find the answer of the question he asked?

If we provide him a tool that he can use with ease to find the answer of his question would be self service in nature in my perspective.

Best example of self service in my mind is Google Search. Do any body need data scientist or analyst to find the answer of his questions on Google? absolutely not. You simply ask your question and Google handles all the complexities, data models,Indexing,performance,mappings etc about my query and return me the best possible answer of my question.

Do we have such an easy interface available in BI universe to find the answer of business man question? A tool that a business man can used to connect to his data and ask  his question "How i can increase my profit? whatever complexity are there to answer this question should be hidden and science should handle it automatically. I will call it a self service BI. Exposing the BI to business!

I am not sure how it would be achieved but that would be true self service BI

I will appreciate your further thoughts on this.

Comment by Homer Roderick on August 7, 2014 at 11:29am

I agree with Michael. It depends on you definition of BI. I like to classify it into tiers

  • intelligence - the capacity to understand or learn which varies by individual and requires motivation
  • knowledge - specific granular truths known or discoverable
  • wisdom - the understanding of relationships like cause/effect and systems

These are pretty broad and general and represent in many ways a person's learning curve or mastery. I start with teaching a motivated client how to use a tool on a simple dataset. They begin answering their own questions for which they've been trained how to do, then quickly advance to answering questions without coaching. Finally they ask for more complex data models be provided and how to do more advanced analysis or presentation,

So really self service depends on and individuals capacity, motivation, and preparation to answer their own questions.

Follow Us


  • Add Videos
  • View All

© 2014   Data Science Central

Badges  |  Report an Issue  |  Terms of Service