Today’s business environment can not understate the importance of business analytics. Most of the businesses leading their industry are leaders in analytics as well. While many Fortune 500 enterprises term analytics as a crucial component in business advantage, it is not too late for an organization to start planning their analytics.
It is already a cliché that CoVID-19 has accelerated the journey to digital. It might already be late if you are waiting for this crisis to pass and then focus on digitization. With crisis impacting day-to-day operations across most businesses, setting up an analytics team when everybody is downsizing and shifting to leaner business models could be termed as daring, if not outlandish foolish.
Most of the frontline and operations teams are intact while axing the value-additive and research teams. In times like this, does investment in analytics makes sense?
Most of the analytics in traditional businesses focus on driving topline growth. Starting with sales performance reports, customer profiling, all the way up to marketing-mix and recommendation models. Think of businesses involved in manufacturing products and selling to end consumers through retail partners- FMCG, apparel, electronics, automotive.
While analytics can definitely deliver revenue growth, the same intelligence when used in automating and improving processes e.g. predictive maintenance, fraud prevention, inventory optimization can lead to cost-cutting as well. Moreover, there has been an increase in digital adoption. This indicates there will be an acute need for analytics for delivering better customer experiences- through chatbots, product personalization, and targeted digital campaigns.
Learnings from The Past
The Worst Precedent in History
Despite widespread adoption of analytics due to surge in popularity and demand in the latter half of the past decade, there’s only a handful of organizations that can boast of a meaningful RoI from their analytics investments. Advances in smartphone adoption, a 4G wave, and open-sourcing of R&D in analytics have led to the democratization of data and technology- ensuring a level playing field for everyone. Still, some industries (think of BFSI, e-commerce, social media) are ahead of the curve, while traditional businesses and developing countries are playing catch-up. With entry barriers no more there, major challenges still lurk in the form of implementation and scaling of analytics initiatives.
Before setting up analytics, there is a need to step back and understand how analytics is going to complement an organization’s capabilities, what roadblocks one needs to sidestep and which transformational exercises one needs to undertake to extract the most value from analytics implementation, and that also, in an efficient, cost-effective and continuous manner.
Roadmap to Business Analytics
The Man In The High Castle
The decision to move towards analytics enabled organization should be backed by top management. And should be accompanied by an organization-wide change in strategy. While chalking out the analytics roadmap, a top-down approach is required with clear ownership and well-defined leadership roles. Commitment and investments need to flow initially which can be mapped to realistic timelines and goals for implementation, adoption, and RoI delivery. A center of excellence – headed by CIO or CDO, taking care of analytical needs across geographies and functional units – needs to be set up which can act as a thinktank for all business decisions.
Priorities need to be assigned to different business use-cases with a clear cost-benefit vision. Each potential opportunity can be scored against expected business value, monetary costs, implementation feasibility, development time, and so on. An assessment of existing analytics maturity in an organization and comparing it with best industry practices can be a good starting point. While everyone wants to chase flashy and tech-savvy AI-based solutions, significant value can be derived even from simpleton solutions like- automation of reporting to ensure that CXOs get their dose of insights in real-time and create dashboards with drill-downs so that business managers can easily identify pain-points.
The One with All The Juggling
With a demand-supply imbalance tilting in favor of analytics professionals, hiring a right mix of talent across business managers, data modeling experts, and data engineers can be a huge bummer in terms of time and monetary costs. For small organizations, it often becomes a choice between small teams- multi-tasking members with shallow skills across the spectrum VS specialized and experienced team members leading to high costs and low productivity. These hiring decisions also have to be made in parallel with a choice of analytical toolkit (Python vs R vs SAS), data infrastructure (SQL vs no-SQL, cloud vs on-premises), data viz tools (open-sourced vs Tableau vs PowerBI)- you don’t want to hire a SAS developer if you are not willing to buy SAS software for USD 10k per annum per user.
It’s also worthy to note that significant analytics tasks like setting up data environment, report layout design, and implementation, creating ML models- might be one-time set-up tasks and after that, automation or minimal maintenance can keep these rolling. Hence, it doesn’t make sense to have large in-house teams for eternity. Further, with the advent of new algorithms, and new platforms popping up every day, you risk being outdated very soon. Hence, we need frequent investments in training teams on cutting-edge tools and techniques. Bringing in an external analytics partner could help in the mitigation of some of these challenges.
While there has been an explosion of data in recent years, data collection and storage practices have not kept up with velocity & volumes of data. With data being there in abundance, every business unit still collects and maintains its data. The result is, we lock away most of the data in silos, with no or relatively low usefulness. Also, most organizations still rely on only legacy data sources. For example sales data and inventory data, which have almost no information around customers, demographics, competitors, manufacturing operations. These data sources along with external data can be valuable in providing insights into customer behavior, process efficiency, market share, and pricing strategy.
Not just the choice of data and data sources, setting-up a dedicated process and infrastructure for data integration and data management is equally crucial- choice of hardware and data warehousing tools basis data type, usage, and volumes. Further, at the forefront of data culture is data privacy and security concerns. There are tighter data regulations (for example, GDPR, in EU) around data privacy in place. Hence, there is more need than ever for ethical data collection and storage practices, anonymization of sensitive data, and user’s consent for data usage and publication.
The following table showcases some common data concerns and suggests the respective solutions-
Data Security and Privacy
Culture Shift Challenges- Changing Status Quo
Most organizations approach analytics as another IT project, which can be taken care of by a separate department- that’s hardly true. The realization of real value from analytics happens when key decision-makers across all functions and all levels practice analytics. There is a need to inculcate data-backed decision making, replacing conventional wisdom and gut reliability. Apart from a general human tendency of resistance to change, this friction is further driven by a lack of proper technical and data analysis skills among business managers. Hence, we lose most of the analytics projects in translation somewhere.
Analytics teams should also share a part of the blame. AI practitioners driven by a hunger for model accuracy and limited success of traditional ML models with big data have increasingly embraced neural networks which are a black box. Hence, they provide no (or at best, a pseudo) explainability for their predictions. And in corporate boardrooms, where each decision could be worth millions of dollars, business leaders tend to rely on their wits and experience than relying on some robotic predictions which they can’t decode. Rather than asking business owners to accept ML models blindfolded, undertake a “root-cause analysis” approach that focuses on underlying factors behind predictions to build much-needed trust in analytics models.
We need proper alignment and co-ordination between multiple departments for the smooth deployment of analytics projects. Analytics can identify risk in credit or fraud in insurance. But it is up to relevant stakeholders to devise actionable strategies from those insights. For example, some short-term questions that need answers are- Do we need to reject such customer applications? Or charge them higher interest/premiums? If yes, how much? In the long term, tweak the customer acquisition strategies to acquire a lesser mix of such customers.
Hence, it becomes important to ensure collaboration and alignment between analytics and functional teams. Right from the identification of a business problem to its translation into analytics problem and evaluation of implementation feasibility.
And Nothing Else Matters
Not just analytics, one needs to gauge any initiative against the actual impact delivered. Only then, investments can be justified and pitfalls, if any, can be mitigated. Roll the projects with a worthwhile business value out to production and scale them across territories. There should be appropriate customizations while those unsuccessful due to tech limitations, data availability or execution restrictions are not to be pursued further. For analytics, this whole exercise assumes more importance. Because any change in the business scenario will lead to a shift in business KPIs and variables causing any historical comparisons or old ML models to be obsolete. Hence, we need to fine-tune the analytics approach and models to make sure that we reflect the latest data patterns and business trends well.
Also, different stakeholders will have different goals and different expectations from analytics. Hence, we need a holistic framework for impact measurement. The following is one such framework spanning across multiple dimensions.
The Way Forward
Most organizations are still at an initial stage of analytics maturity and the force of digitization in full swing. Hence, the best of analytics is yet to come. An organization’s ability to successfully transform itself into a data-driven enterprise is going to be a key differentiator between leaders and laggards. While data & AI may not replace business minds and human intelligence, it can be a valuable strategic asset and key competitive advantage for businesses. Transitioning to data-driven enterprise, a choice now could be very soon a necessity for survival. To make all this happen, there is a need to rethink one’s analytics approach to enable value discovery, opportunity identification and prioritization, and widespread adoption at the grass-roots level. Needless to say, this intent has to be backed by much-needed strategy, leadership backing, and investments.
About the Author
Data Science and Artificial Intelligence professional with 15+ years of experience across industries. Have been on both Company (Vodafone, Aviva Insurance, GE) and Consulting side. A passionate advocate of data science and have a constant endeavor to create optimal actionable solutions to derive measurable business value from data. Currently, Director – Intelligent Automation and Accelerated Analytics (IA3) at Nexdigm Business Consulting.
Connect with Amit – LinkedIn