Data ecosystem – Foundation for robust analytics and AI

We are seeing renewed interest in data platform discussions in recent time. The concepts like CDP (Customer Data Platform) are at the forefront, the place which was occupied by AI (Artificial Intelligence) and ML (Machine Learning) until sometime ago. Those were the times when businesses realized that the new age artificial intelligence and machine learning processes could help in solving many challenges and can add significant value. And everyone wanted not to be missed in this AI/ML bandwagon. Many of them took significant steps and invested resources to reap the benefits of these advanced techniques, but when they reached to the stage of detail deliberation in order to make those fancies a reality, they realized (or rather re-realized) that AI and ML (and anything of this sort, even simpler) are just some pieces of the puzzle. For AI, and for analytics in general, there is something downstream – data, and upstream – operationalization, which needed to be considered together with AI/ML strategy (I will refer this as analytics instead). I used a circular construct to represent this in my earlier blog ‘The trust cycle: Personalize like you mean it‘. I mentioned, in the context of improving customer’s trust in brands, that there are three possible intervention points for businesses – Data, Insights, Engagement.  

In this blog, I further developed that construct to a framework of three integrated ecosystems. In the illustrative image, data, analytics and customer engagement are parts of this framework of integrated ecosystems. And data plays a pivotal role as foundation of this integrated framework.

The data ecosystem is responsible for providing single view of customer to the analytical platform for generating further insights about customer segments, references and predict future behavior. The single view of customer, along with the insights generated by the analytical platform, drives an effective customer engagement system. Moreover, this isn't a one-way process, the data being collected from the customer engagement system again increment the existing customer information. For this whole ecosystem to work efficiently and holistically, these three eco-systems should function in an integrated manner.

Thinking data ecosystem – CDP way

Let us look at this data ecosystem through the lenses of established CDP capabilities and explore what would be the high-level questions to ask to evaluate the preparedness of the data platform or ecosystem:

Data ingestion – Businesses typically collect data at separate places through variety of systems – mostly because of the technologies used to collect them. E.g., CRM interactions are collected at CRM systems, but the digital behavior is collected somewhere else, typically in SaaS cloud. Businesses often also subscribe to external agencies and systems to gather data which they generally don’t have access to. The question here is: Do we collect all we need? Maybe we should instead ask: Do we collect all of what we can? Because many a time our needs are shaped up based on what we have access to.

Integrate data - It is often said that big enterprises usually have lots of ‘unused’ information about their customers. One major factor behind that is fragmented and disintegrated data. This is quite challenging to create a single version of customer record by breaking the data silos and legacy system barriers. The questions here are: Are all available data sources appropriately integrated? Followed by: What are the latencies, can they be updated in real-time to the extent possible? 

Provisioning data – The integrated data is of no value by itself until it can be accessed by the systems who can make use of it. The possibility for other systems to access and use this data – on demand, in real-time, etc. is critical. We would ask: Can my frontend and/or other operational systems access data when they need? 

Don’t get me wrong – Analytics, AI and ML are critical

The fact that the data ecosystem is foundation, doesn’t undermine the importance of analytics, AI or ML. Instead, in an integrated framework, an optimal and functional data ecosystem will enhance the value and effectiveness of the analytics functions. In my previous blog ‘Boost your bank’s operational efficiency’ I discussed about how a central analytics hub could help even in breaking departmental or functional silos.

Operationalization or deployment of analytics into the operational processes is another factor. The construct suggested above takes example of customer engagement as one (out of many) operational area. It is very important aspect to be considered while planning the project. A research by Venturebeat states that about 87% of data science projects never make it into production. I mentioned in my previous blogs the importance of operationalization of analytics emphasizing analytics should be an integral part of business processes. (‘Why customer journey design matters (and having analytics at its heart matters, too)’ and ‘Operationalizing retail analytics to accelerate addressing new customer “sub” segments’)

It is truly said… Data is the new oil..