Research2
BI Front Ends, the “Face” of Analytics
1
What’s Our Focus This Week?
2
Back End
(Storage)
Extract, Transformation and Load
(ETL)
Front End
(Projection)
+
+
Self Service BI = Nirvana
No IT involvement, users are empowered to access and analyze data themselves
No pipeline of requests, no delay
Change occurs at the pace of business
The reality is different due to staffing, complexity and skill level
3
Source: Howson
Self Service BI
Data Scientists
“Play” Here
IT Creates
Semantic
Layer
IT Creates
Semantic
Layer
IT Develops
IT Develops
4
Source: Howson
Business Query and Reporting Tools = Ad Hoc Query Tools = Self Service Access
Allow business people to create queries without having to know SQL
Business user builds the report, not IT
Business can’t wait, so they
Either demand their own environment supported by IT or
“Go it alone”
True ad hoc = a one time effort. What are often called “ad hocs” by the business are really production reports they can’t live without
BI is iterative so these “front end” display projects tend to be Agile while “back end” data loading and cleansing projects tend to be waterfall (more on this later in the course)
5
Source: Howson
What Makes A Report? Formatting…
Standard reporting templates are important for corporate look and feel
Font style, conditional formatting
Cross tab, chart, master – detail
Complex: multiple charts on a page
6
Comparison: Predefined and Ad Hoc Reports
7
Source: Howson
The Semantic Layer = The Business View
Uses business terminology vs. physical field names
Automatically connects related tables via SQL joins
Provides metrics that calculate and aggregate facts
Up front investment yields good ROI:
Consistency
Lower maintenance
Higher accuracy
8
Building the Semantic Layer and Reporting
Step 1: Import the Data Structure
Step 2: Design the classes and
objects:
Step 3: Implement the semantic
Layer
Step 4: Users build queries with the
query panel
Step 5: Users generate reports
(users from IT or business)
Source: SAP Business Objects
Note: These screens show
the development flow
and are not related
9
Challenges
Since BI tools are the “face” of BI they get an inordinate amount of attention. Users see the tools so they will blame them for any problems
Ad hoc reports developed by the business tend to be more robust and effective than those developed by IT. Unfortunately, IT doesn’t know about them!
Most of the time, IT leaves it to the business to test “ad hocs”
Good idea: Open dialog to “productionize” the business’ “ad hocs”
Without a clear strategy, static reports become redundant with both themselves and ad hoc reports
Consider a report rationalization effort to consolidate and achieve a single version of the truth for each predefined report. Publicize this with the business to save on their “ad hoc” production reports
When users “go it alone” they usually run into scalability issues and IT is pressed into service at the 13th hour to take on support
10
Challenges (Continued)
Make sure you design a semantic layer that is intuitive to the business community (surprisingly, this often does not occur!). If not you risk:
Rejection
Inaccuracy
Perception of failure
11
Visual Data Discovery
This niche is growing at three times the pace of the overall BI market. It is evolving and not yet well defined
Visual data discovery is “…a tool that speeds the time to insight through the use of visualizations, best practices in visual perception and easy exploration (Howson, 2014)”
Users may not know what they’re looking for until they navigate and drill within a data set for details and trends
Visual discovery tools may lack the business metadata layer and tabular data sets characteristic of business query tools
Provides agility and ease of data access and insight
12
Source: Howson
Comparison: Business Query and Visual Data Discovery Tools
13
Source: Howson
Example: Tableau Visualizations
Dashboards
Like car dashboards
Present information from multiple data sources
Present multiple numbers in different ways
Include highly visual indicators and/or reports
Business users want to customize their own dashboards with relevant information
Some dashboards, like those sourcing operational information directly, require IT support
15
Source: Howson
Dashboard Example
Production Reporting
Static reports with sophisticated formatting and design
Also known as:
Pixel perfect reporting
Operational reporting
Enterprise reporting
Canned reporting
Predefined Reporting
May access operational systems directly, an operational data store or detailed data in a data warehouse
Developed for:
Avoiding “run away” queries
Embedding within a transaction system (i.e. inventory levels, bill of materials, invoices, etc.)
Presenting common enterprise requirements
Static reporting (Example: Reinsurance statutory reporting)
Management reports by IT in lieu of available production reporting tools
17
Source: Howson
Example: A Bad Production Report
18
Comparison: Production Reporting and Business Query Tools
19
Source: Howson
The Case of the Production Report vs. Business Query Debate
IT recommended batched production reporting for preprocessing and availability via SAP Crystal Reports
Driven by concerns about data volumes and load times
Users were accustomed to static reports and IT concluded they still wanted that
With predefined reports, each report would needed to be developed, replicated, then maintained
Users then requested the templates used to create these production reports to be made available for querying and ad hoc report (i.e. “shadow production report”) development, so SAP’s Business Objects Web Intelligence was used
Concerns about “pixel perfect” were addressed: both tools satisfied the need
Concerns about rendering and data load were not resolved, though preprocessing was available via the Business Objects administrator
Semantic layers could be shared
A huge, extended dialog ensued where we decided to create and reuse templates (Prof saw that coming on the first day of the discussion)
20
The Argument
21
Source: Howson
Mobile BI
Similar to the internet, mobile is ubiquitous, someday it won’t be “special”
For now, the mobile BI challenge remains unique:
Design for small and varying screen sizes
Bring Your Own Device (BYOD) introduces managerial and security challenges
The user experience with browser based access is not optimal, native device-based apps should be used for accessing, authoring and viewing content
22
Source: Howson
BI and Microsoft Office
Unofficially the leading BI tool (good for Microsoft!)
Wreaks havoc on BI’s goal of providing a single version of the truth
The preferred user interface
Need to facilitate integration while managing use
Old: Export from a managed environment to an Excel spreadsheet. The result was chaos
New: Most vendors integrate with Excel to extend BI’s reach
Office 2016 includes visualization and in-memory processing
Exporting data to Excel spreadsheets is not BI and is dangerous, managing Excel use to explore data is beneficial
23
Source: Howson
Reference List
Howson, C. (2014). Successful business intelligence: Unlock the value of BI and big data. New York. McGraw
Hill Education. ISBN: 9780071809184
24
