Decision Mashups

November 7, 2009
68 Views

Sunlight_example Last week I presented my thoughts on mashups at the Data Warehouse 2.0 Asia Pacific Summit. I talked mainly about why mashups can be a useful item in your toolbox of analytic solutions and looked at some best practices.

Why Are Mashups Popular?

  • They are less technical (no programming skills needed)
  • Data is easier to get (XML)
  • Free data tools (Google, MSFT, Intel, etc.)
  • Bandwidth is available
  • Web 2.0 (Internet as an interactive platform)

What Are The Benefits of Mashups?

They deliver a greater variety of:

  • Data to decision makers without having to (formally) invest in data integration and warehousing
  • Presentation options so that the data is better understood (greater insight).

Can enhance trust:

  • If you can access data from another internal data source/warehouse that is already established
  • If the external data is from government, regulatory or other trusted source.

Helps people think about data and logic (information centric)

Increases the potential size (and value) of the user community (many ‘programmers’)

Can deliver extra value to customers and partners (analytic suites)

Enhances your analytic community (encourages cooperation)

Encourages innovation (unintended uses, knowledge worker can react ..



Sunlight_example Last week I presented my thoughts on mashups at the Data Warehouse 2.0 Asia Pacific Summit. I talked mainly about why mashups can be a useful item in your toolbox of analytic solutions and looked at some best practices.

Why Are Mashups Popular?

  • They are less technical (no programming skills needed)
  • Data is easier to get (XML)
  • Free data tools (Google, MSFT, Intel, etc.)
  • Bandwidth is available
  • Web 2.0 (Internet as an interactive platform)

What Are The Benefits of Mashups?

They deliver a greater variety of:

  • Data to decision makers without having to (formally) invest in data integration and warehousing
  • Presentation options so that the data is better understood (greater insight).

Can enhance trust:

  • If you can access data from another internal data source/warehouse that is already established
  • If the external data is from government, regulatory or other trusted source.

Helps people think about data and logic (information centric)

Increases the potential size (and value) of the user community (many ‘programmers’)

Can deliver extra value to customers and partners (analytic suites)

Enhances your analytic community (encourages cooperation)

Encourages innovation (unintended uses, knowledge worker can react quickly to the business)

Changes the type of innovations (pull by the BI function becomes push from the business)

Can offer better data quality (integrity, completeness and freshness)

Saves time (faster development: days not months, ROI, lowers barrier to entry)

Low infrastructure costs (individual computers as servers)

Some Potential Issues

C level support You are using data sets that haven’t been conformed to the same standard as your static data stores (DWH, ODS, etc.)

Users will see the quality of data (little ETL invested, master data not used)

These data sets may not be:

  • Always available
  • Consistent over time
  • Refreshed consistently

Access security

Expectations management (“why can’t I see all my spreadsheets?”)

Application performance may suffer (inefficient queries, network traffic spikes)

Tactical developments may overtake the strategy

If done in a self-service environment:

  • Standards
  • Maintenance (support, documentation, ownership)
  • Scalability 
  • Availability

may all suffer (lose IT SDLC benefits)

Resource misallocation (budgets?) Who governs use? (multiple SSOT)

Here is a (13MB) pdf of the presentation: Decision Mashups: Strategies for Combining Structured and Unstructured Data for Better Decision Making. Please note that I have removed a couple of case studies to reduce the size of the file.


Link to original post