10 Parameters for Big Data Assessment

Organizations are gradually thinking of getting more value from their big data solutions and the underlying datasets. These datasets provide the opportunity for organizations to deepen business insights and predict scenarios that will further help them to stay relevant and prominent in this competitive world. Many of the organizations across the world are trying to assess their current big data and analytics environment so that they identify performance bottlenecks, how optimal the solution is and how equipped is it to cater toda new requirements. The assessment is broadly based on two things.

1. The existing environment with current requirements

2. The existing environment with Future requirements

The existing environment with current requirement: It spells how successful the current environment is by validating all the implementation of previous requirements. It means to what extent the current analytics and BI satisfy the needs of the business. The current solution needs to be validated against all the 10 parameters as mentioned in Illustration 1.

It is highly unlikely that the current solution holds its fort against all the parameters. Some of the typical issues around the solution are;

1. It may not satisfy all the analytical needs of the business and some of them may require significant changes as proposed by the business before it can be used

2. The performance of the execution of analytics is not up to the mark

3. Many loopholes in the security of datasets as anyone within an organization can access any part of the information

4. The business doesn't treat the solution as a single version of truth because of data quality issues

5. Business need to wait for a longer duration before the new requirement can be materialized

6. The master data which is a reference for many analytics can not be trusted because of duplicates and non-standardization of data

7. The data set which is a feed from one of the newest data sources cannot be accommodated because of non-flexibility of the solution

8. Business feels that they have to wait endlessly before any of their new requirements are implemented

The demands of the business manifold as it undergoes changes over the period of time due to new acquisitions, new business models, campaigns and new structure. However, the ill-designed big data solution may fail on many of the above-mentioned parameters (illustration 1). It is vital for any organization to find big data issues and bottlenecks before any major failure occurs. This warrants the assessment of big data solution so that all the mentioned parameters are assessed to the deepest level and identify all the issues that need a fix. The assessment can be done as per the below phases.

The planning phase will primarily define the scope of the assessment with specific timelines so that all the tasks can be tracked. It is recommended to initiate the entire plan using an agile framework so that periodic checks can be done.  The discovery phase will conceptualize various layers of the solution for bottlenecks and brainstorm with various users so that more clarity can be obtained. Based on the investigation and findings, a recommendation document would be created and socialized with all the stakeholders. This will lead to a roadmap of implementation which can be categorized into 3 stages;

1. Immediate fixes and tuning (within 1 to 3 months)

2. Midterm implementation (within 6 months)

3. Long term implementation (from 6 months to 2 years/more)

Questionnaire

The assessment questionnaire plays an important role in understanding the current environment and bottlenecks. The assessment will not achieve success without a well-designed and drafted questionnaire. The mode of the questionnaire will depend on the requirement of information and it may be factual (objective) or attitude (subjective). The initial set of questions can be of attitude however as the schedule progresses, it should be more of factual questions. The factual questions reduce the ambiguity and help to build the information around the challenges and issues that the existing solution consists of. The questions should be grouped as per illustration 1 so that all the sections of the solution can be assessed.

Assessment Score

The assessment of various sections as illustrated gathers a lot of data that will have responses for both current and futuristic needs. Along with the needs, it will also have the current issues and bottlenecks that need to be addressed either in the current or futuristic solution. This gathered data needs to be presented to the organization so that a timely decision can be made. The best way to represent the assessment is with a scorecard. The scorecard will have the numbers reading from 1 to 5 and the assessed information will be tagged to one of these numbers based on the current level. For example, the data quality in the current solution is assessed at 3 out of 5.

The assessment score can be generated using many algorithms however using the right set of parameters is vital to generate an accurate score. The parameters are;

•  As-is

•  To be state

•  Confidence

•  Risk level

The above-mentioned factors ensure the right score highlighting the areas which needed immediate attention where the risk/impact is more.

If any queries or comments, please write to basu.darawan@gmail.com