Analytics - How to know what to implement?

The capture and usage of data is forcing organisations to think about short and long term technology transformations. By using a heady combination of apps and connected devices, managers use "network on the cloud" concepts to deliver state of the art analytical capabilities. These tools help people to clearly think beyond the regular "SQL reporting tools" to understand what a customer needs.

Businesses now think about the needs of an individual user. These days, an organisation's vision would be to substantiate sales by using the massive amount of data of that customer, which is collected from a wide range of primary and secondary sources. And what's more, they now want to use the cloud to deliver this capability for them.

While the needs of the organisation always sound exciting, their requirements are often specific to themselves and do not have much of a solution re-use angle, even within their industry and domain. Adding to all this conundrum, organisations now face some cost related questions as well, since an analytics program does not come cheap. Som organisations do pose some challenging questions - How can I implement an analytics solution? Does it have to be an independent program, or can it be embedded into my next application project? Which team do I first roll it out to? Can my application vendor deliver a BI program as well?

The short answer is...it depends on your use-case!!!!

The last solution I helped deliver for a retailer, the focus was to avoid the "swivel chair" which is different teams looking at different applications to get the data they need. This makes the user waste productive time. Also, the organisation did not just want to look at purchase history, their definition of valuable data included social behaviour, website analytics, community feedback/comments and even big-data trends. This is a marked difference in trend from some of my earlier projects, where the focus was to create reports based on consumer financial status data, demographics and even age, most of which is something often collected from a customer base.

While both the above projects were delivered using cloud solutions, the role of the business user in both the programs ranged from being a regular "data consumer" to being an "data analyst". While the data consumer prefers to answer some filtered questions to look at a predefined data asset, the analyst prefers to mine data to share with the larger organisation, say management/sales teams. A data consumer will usually need a simple QnA or a drag and drop tool, whereas an analyst will have specific training and experience to use OLAP tools which is often complex for the normal user.

So how do we look at this challenge? How to know if my organisation has consumers or analysts or both? I have collated a few parameters to look for while arriving at an approach for solving the analytics problem:

• Organisations would need a thorough understanding of how data is persevered across systems. This helps us understand a host of issues, including structure, interoperability of systems and confidentiality of specific attributes. This may also help us look at the way this data is being stored, the location of data and how easily this can be recovered for use.
• Who and how is this data more valuable for. This determines user experience (UX) associated with fetching and preparing data, and drives adoption
• Find out ways to integrate analytics to the enterprise, including the fact that some of this need to be embedded into an application's workflow. Also, it is better to address the "do you really need analytics for this problem?" first up rather than later.
• Who needs to see what data, i.e. security, with a view of who needs more depth of information and who would be fine with just a high level view. This helps us to ultimately answer how a single analytics tool addresses future challenges.
• Analyse On-premise vs cloud analytics approaches, since the latter has been a preferred choice lately. This is also dependent on the data perseverance question as well as the storage costs angle.

From my experience, the approach for resolving a critical analytics challenge lies in the use of a single tool (not many tools, since they typically confuse the business user) that is deeply rooted within the enterprise architecture. Visualisations around these tools could be then streamlined for business teams to deliver the data trends they look for.

The focus, at the end, is to deliver a solution that can transform the organisation to use approaches where analytics can be seamlessly embedded in all its key applications. This drives an approach to deliver a product with rational and sustainable features that puts organisations ahead of its competitors.