Steps to getting right your data analytics effort

Data is growing at an unprecedented velocity. This holds firm ground in the context of today’s digital age fueled by the increasing use of IoT devices and sensors, software as-a/for medical devices, AI-powered health applications for monitoring and tracking, e-retailing, use of chatbots, digital identities, growing use of social media for peer consulting over the internet, data processed/stored in databases, etc.?

Traditionally and mostly, the data analysis revolves mainly around displaying statistics and trends which is talking about what happened in the past. As this data builds up over time, traditional tools and systems struggle to handle huge data volume and process it efficiently. The advancement of technologies and the option for off the shelf tools today relatively makes it easier to look beyond what happened in the past to what could the future look like.

Data analytics/science today is imperative for organizations to tap into the source of differentiation. Combined with AI / ML it could help generate insights that could be used to achieve operational excellence, enhance customer experience or improve/innovate the product /service.

Getting your data analytics effort right

1.??????Build KPI's anchoring around the problem or target

Define the problem/pain point you want to alleviate or areas you want to improve or have better control. Build meaningful KPIs anchoring around the problem that will help you monitor and ascertain the health of your processes or outcome. Your KPIs will roll up to the objectives you wish to achieve. Keep the feedback loop open so you know if you have to be flexible with targets or results in your journey to achieving objectives.

2.????????You will need new investment- infrastructure, tools, etc.

Your existing infrastructure would have been built for a purpose a long time back and it may not meet the growth capacity and agility needs of the modern operations required for analytics e.g., for a data warehouse, and running analytics tools and operations. You may also need to develop APIs to pull data from the databases, or from varied sources e.g. email, forms, social media, mobile applications data, etc.

Work with your vendor/partner to design the infrastructure needs. Purpose-ready cloud environments (PaaS) for big data processing are available with key cloud service providers. Leveraging cloud PaaS would not only reduce your upfront spending but also reduce the lead time to experiment and start yielding results.

Remember, data analytics is an iterative exercise. Model creation (analytics report) is one aspect of the operations but it is equally important to have a business sense for the data engineer to model the right analytics template.

3.??????Build data analytics competency

The raw data needs to be curated, templatized for use, and analyzed before it churns out actionable information. Remember data analytics is 80% preparing data for analysis, and managing the data warehouse. Remainder 20% is the analysis and science. This is significant work and running this program through your existing in-house IT team may not do the justice, both in terms of time and skill.

Data analytics requires specialized skills that go beyond the traditional IT skills of running and maintaining systems and applications.

4.??????Know and map your data landscape to requirements

Chances are bright that the data you need for tracking KPIs could be spread across different functions. This could be especially more probable for B2C companies but also largely determined by what are you trying to accomplish. Challenges arise if these functions operate in siloes (which mostly is the case in regulated industries and not sharing information freely) and in different formats. It is important that your design/architecture documents clearly outline the data requirements. It should also specify the SPOC responsible for providing the data, the format, and the frequency the data needs to be shared.?

Make no mistake or leave ambiguity on the quality and consistency of the data you require for analysis. In the absence of quality data, the analysis outcome wouldn't show the correct picture and would be either skewed or inaccurate thus jeopardizing the entire effort and even initiative to appear ineffective and failing.

Using AI/ML to work on the data would bring effectiveness and efficiency to your data analytics effort by automating specific tasks and actions. Without quality data, AI/ML will serve no purpose.

5. Define the data governance structure/framework

Putting in place an effective data governance framework is a must in today's regulated landscape especially for organizations in the healthcare and life sciences verticals that deal with sensitive health data. The data governance model should talk about the entire data's life cycle, its classification and purpose, its ownership, its availability, and accessibility needs,

6.??????Gather Sponsorship support

Sponsorship support will help you navigate through organizational swim lanes more easily. Apart from the investment, you will need for the initiative it will provide you the road to run your car. It can help dissolve invisible obstructions across the organization and within the function and garner the needed support. Build a viable and attractive business case that triggers their interest.

Please don't get fixated on efficiency when beginning on this journey. Efficiency would be the by-product of the overall initiative that you will undertake. Keeping your focus on pain points/KPIs is crucial. Please note that the efficiency is inward-facing. For you to create a source of competitive differentiation your analytics should be outward-facing.

7. As people say, the Execution Approach matter more

Gone are the days for an executive to decide between build vs buy. Given the options available, building it from scratch is laborious, time-consuming and an uphill task that can eat up your precious lead time to market and delay the differentiation opportunities you may have. The choice remains whether with the options available on the shelf you want to drive and do it yourself with vendor support or use a service provider or partner model to let them do it for you. Consider factors such as the uniqueness of work or experimentation intent, the scope of work/vision, and the skills available at your disposal while deciding on an execution approach. ?

Most likely than not the CIO may be asked to lead the initiative given the amount of technology and infrastructure work this initiative will entail. Avoid making this initiative a technology project and business being a bystander. It's not just technology here but an improvement in business process and approach that we are eyeing which will require breaking organizational siloes that have been operating for years and focusing on their own areas. Your team has to be cross-functional.