Packaged analytics: Accelerate time to value
By : Dr. Prateek Kapadia
Chief Technology Officer
As the age of digital economy dawns upon us, the business growth of enterprises will depend upon their ability to rapidly define, transform or realign their business models as per the evolving market dynamics. Modern day enterprises, in their quest to obtain a customer-centric perspective of their businesses, are incessantly amassing vast quantities of heterogeneous data from an increasing array of data sources from both within and outside the organization. However, as the Gartner Research submits, the ability to turn this massive data into business value remains a huge challenge.
Only the agile will survive and thrive in digital economy
Conventionally, business leaders have been accountable to seek new revenue streams and protect existing ones using data and analytics, while the IT leaders shoulder the weight of the technical development, deployment and delivery of analytical models. There is lag spanning several weeks between opportunity identification by the business team to the delivery of use-case specific analytics model by the IT team resulting into either a missed opportunity or a sub-optimal result.
Time to value through conventional and packaged analytics approach
This is where packaged analytics as a concept becomes relevant. Enterprises need to take a value-centric approach to analytics. Analytical models have higher inherent value, if they can be readily deployed in a matter of days and can allow business users to carry out further business specific refinements quickly, before being rolled out for commercial use. Packaged analytics is all about ensuring availability of such ‘best fit’ and ‘high performance’ analytical models for enterprises in different business contexts.
Innovate early and innovate often
With packaged analytics, business leaders can aspire to be among the elite 10% who can quantify and unlock the economic value from data by launching innovative use-cases at the right time. It will also allow enterprises to deploy advanced analytics on a small scale for a single department or single application and then expand seamlessly to support other departments and applications using the same model and platform. Deploy quickly, experiment iteratively, and scale rapidly – is how enterprises need to think about putting in place analytics capabilities and that is what packaged analytics promises to deliver.
Packaged analytics need to be seen as a modern business-user-centric capability. Its success has to be measured in terms of ease and speed at which enterprises can deploy and execute the models as well as iteratively refine them based on the model output.With pre-built integration with various data-sources, issue-specific KPIs, machine learning framework, analytics workbench for experimentation, and best-practice templates, packaged analytics can help in accelerated adoption of analytics across the organisation.
Packaged analytics models need to integrate all the components required to deliver a ready to use solution:
- Connectors for data ingestion from multiple data sources (Real-time, Near-real-time, Batch)
- Data preparation tools (ETL, Anonymisation, Aggregation)
- Pre-defined KPIs for specific business problems
- Machine learning framework and analytical algorithms
- Analytics workbench for fine-tuning, performance enhancement
- Self-service visualisation dashboards
Accelerate time from data to decision to dollars
Now let us see what promises packaged analytics hold for enterprises and business users. It will help them to do faster, deeper, objective-driven analytics more efficiently with a self-serve interface, significantly reducing the ‘data to decision to dollar’ time.
Telecom industry is among the frontrunners in generating a humongous volume of numerous varieties of data. With Packaged analytics telcos can embrace this new reality to drive operational efficiencies, embark upon new revenue streams, and improve the customer experience through data driven insights. A snippet of economic value delivered to various telcos through packaged analytics models:
Examples: packaged analytical models
In addition to the packaged analytics models that drive the internal monetisation for telcos with subscriber insights such as churn score, retention score, channel affinity, next best offer propensity etc., there are advanced analytics models to open up external monetisation revenue streams as well. For mobile advertising consumer insights such as E-commerce users, Mobile-wallet users, Rural-consumers, International travellers etc. can enable brands to engage with the most relevant audience through an optimal channel. With Packaged analytics Telcos can also deliver useful insights to partners in the digital economy such as device migration patterns, brand loyalty score, population density etc. to drive additional revenue.
Packaged analytics approach: From inception to action
Flytxt offers 100+ proprietary packaged analytics models of which 80+ models are running live with 7 telcos across 5 countries, consistently delivering 2-7% of economic value over last few years.
Having established the value of packaged analytics, now let’s take a closer look on how to develop a packaged analytics framework that can deliver ‘right fit’ and ‘high performance’ analytical models to enterprises.
Flytxt has a cross functional team including data scientists, developers and testers to build, customise, package, publish and maintain a set of packaged analytics models in the model library. The objective of this team is to build models to provide a repeatable and consistent solution to a wide variety of business problems using industry best practices and techniques.This enables faster rollout and empowers Dev-ops to instantiate and monitor models across various clients. The platform is then designed to allow the clients to subscribe and experiment with the models from model library and offers run-time environments for executing both data processing scripts and analytics models. Here again flexibility of platform to support different technologies and scripts is the key.
Flytxt packaged analytics delivery framework
The fast mover advantage with packaged analytics
With Packaged analytics, enterprises are equipped to embed intelligence into business processes, operate at transactional speeds, deliver insights to customer-facing employees, and provide ever-deeper insights for decision making.
Packaged analytics offers an opportunity for data and analytics leaders to simplify the analytics landscape and reduce IT complexity with a self-serve capability for business users to accelerate time to insight and improve business value. As the concept and delivery framework matures, we will begin to see accelerated adoption of packaged analytics across the enterprises.