Wednesday, April 5, 2017

Delivering business value using analytics & technology without worrying about maturity models

Delivering Value Through Analytics
Organizations across industries are in the process of increasing digital engagement, closely integrated with analytics, for their candidates, employees and customers. When we meet organisations attempting to do bigger and better things, we often turn to the maturity model concept, both as a means of diagnosing the level of the firm’s capabilities in a given area, and of explaining how that capability can be enhanced to ever greater heights.

Maturity models describe the characteristics of maturity across a multi-point scale, from zero or one (typically a chaotic state), through to say five, where the capability is at the highest level of maturity and creates a genuine point of difference for the firm. For analytics, the journey has many layers and it pushes from transactional reporting of lag metrics to proactive reporting of lead measures to prediction and prescriptive analytics/insights in various business scenarios.











{Circle denotes what % of organizations are at what stage of analytics}
Source: Bersin by Deloitte | DUPress.com             

The Myth of Maturity Models
In the domain of advanced analytics and big data, large analytics consulting providers or core analytic firms use a variety of factors to pin-point a firm on the analytics maturity scale:










Let us first break the myths of traditional maturity models and share learnings from not so successful digital transformation & analytics journeys of mid to large size organisations.

Let me share with you a couple of examples from my research, observations and interactions amongst large IT services firms in India*. 
Example 1:
A top multinational IT services firm with large offshore presence took a bold step a few years ago to automate descriptive business and HR reporting completely through Qlik, assuming it would be able to expedite decision making to positively impact business & HR metrics. In the third year of its journey, full-scale execution of all business and HR reports is still on with very little impact on decision making. Full credit to the organisation for reducing reporting TAT (Click of button vs. reactive) and reporting & analytics team optimisation saving dollars. However, this is a notional saving given the investment made into vendor evaluation time, leadership time & cost and internal team deployment time & cost which would easily take another 2 to 3 years to recover.
Challenges/learnings from the long execution cycle:

Technology integration capabilities (Synergies of buyer & vendor systems)
Vendor domain knowledge for faster transition/automation
100% reliance on reporting automation can stall short term benefits of semi-automated descriptive reports

Example 2:
Another large multinational IT & consulting services firm looked to optimise talent acquisition cost & time through advanced algorithms and analytics three years back. The firm started the journey by automating candidate screening process through NLP & artificial intelligence technology platform, followed by integrating predictive analytics to score candidates on employability/fitment, forecast probability of offer accept to even candidate performance/stickiness through Big Data.
The myth with which the organisation went ahead was the fact that being a PCMM level 5 company, their hiring processes, candidate data, job descriptions, hiring manager behaviour and historical employee data for like-to-like profiles was structured, clean and fully reliable. The pilot and actual launch showcased that while candidate data was rich, the internal data to derive key insights from predictive analytics was extremely weak. As a result, benefits were reaped only on auto screening.

 Challenges and learnings from the pilot & execution experience:
Capabilities – Running a pilot to test predictive analytics capabilities on small scale (Capabilities of not only vendor but buyer data quality, data environment & availability. Any variables with less than 30% data availability cannot be used for any predictive or prescriptive analytics)
The vendor failed to assess descriptive reporting in buyer organisation (Organisations without matured descriptive reporting/analytics capabilities find it difficult to move to predictive analytics)
Cutting edge AI technology with weak data yields little or no impact
It is necessary to issue a caveat here: The maturity model of your analytics practice needs to marry the organisational maturity level. Key questions to consider here are:
Is there leadership buy-in for prescriptive actions from predictive analytics?
Are there sufficient processes in place to capture clean data?
Does organization have a central data warehouse?
Do we have technology to support advanced data analytics?
Does organization have data required to accurately predict?
Leadership buy-in is comparatively easier in financial analytics where we deal with numbers and predict numbers; whereas, it’s an uphill task when the management problem revolves around human behaviour and the attitude ranges from completely predictable to absolutely undecipherable!
Also, it is important to understand that Statistics is only a means to an end which prevents a helter-skelter approach to service delivery. Without doing predictive modelling and creating causation frameworks, there’s still much value that can be delivered to business. A leap in the analytics value chain without the same lever in the organizational maturity would render the outcome useless. 

Here are a few examples to demonstrate the value delivered at different maturity levels without the constrains of sticking to gradual movement up analytics value chain:
AirBnB and Indian bed-linen industry – Nearly a decade back, a handful of financial analysts supporting HNI’s ran simple descriptive & correlation analytics (graphs and dashboards) on growth of bed-linen industry in India and merged with causation analytics to identify possible indirect sources of demands for listed bed-linen companies. This sort of data can be a goldmine for decisions about community growth, product development, and resource prioritization. This led to a conclusion that growth in the number of rooms listed on AirBnB by home owners and the push towards cheaper good alternatives to the established hotel industry would push the demand for bed-linens from manufacturers in India. Five years down the line, both listed and unlisted players have seen a huge surge in both top line and bottom line and decade down the line few players become multibaggers.
Financial Analytics for end HNI customerThe financial services industry in developed nations are truly leveraging the power of prescriptive analytics, providing prescriptive actions to customers on banking and investment decisions, based on factors such as personal health, financial status, weather conditions in cities they live in, political or ecological changes and many more such direct derived data with financial goal set.
Prescriptive Analytics and Consulting for Governments – Most matured countries who have connected systems, high end technology and citizen data captured over years are, in fact, in a better position to leverage prescriptive analytics in policy, political and economy related decisions. Virtual Singapore is one of the best examples where policy makers, statesmen and citizens can build or test decisions through use of IoT and prescriptive analytics. The US government has a team of data scientists testing various decisions to outcome of various policies implemented around the world to help enhance the decision making process. This is a classic example of starting with descriptive reporting to descriptive insights generation to slowly move towards predictive which is extremely mature & requires huge investments.

How to Derive Value?
Value comes from things that matter at that point in time and for the overall organisational purpose. Almost 80% of the reporting done in most organizations is transactional with metrics that are of little value —that’s where we should start.
The right metrics provide a context around which performance can be analysed. With the wrong metrics, an executive summary gets reduced to an anecdotal commentary. One of the classic examples of business metric in HR: Employee Attrition Reporting, Prediction to Prescription. Reporting employee attrition by various bands/locations/skills etc. renders little or no value unless key employee clusters/groups to be retained are identified, reported, predicted and prescriptive actions taken on this cluster to impact employee attrition as a metric.
Such outcomes presented in a compelling fashion win over the management’s trust and get them to rethink processes and workflows, invest in technology and infrastructure, and adapt to the organisational changes that arise out of your analytical frameworks. And when the organisational and business needs are in synergy, you move up the analytical value chain, leaving the management to answer just one question, Are we aware of the next big thing?
Analytics team needs to delivery descriptive & predictive models with Intelligence, Insights and Prescriptive actions to drive value and not just good looking dashboards or statistical models. This is where a combination of statisticians, data scientists and domain experts can together add value to answer the question of next big thing?
Important for Analytics teams to focus on Results but imperative for leadership to be ready to convert prescriptive insights to actions
Significant untapped value lies in data that already exists in most organizations, and analytics team needs to assess required capabilities that can effectively exploit this data. However both leadership team and analytics team needs to establish with a clear focus on tangible business value.

Based on research one of the key reasons for failure of internal analytics team in large organizations across various sectors starting from Pharma, Banking, Financial Services to IT has been lack of synergies and underlying intent of setting up analytics team between leadership & analytics team.
Analytics team needs to establishing an Agile Analytics data architecture and methodology to addresses ever-changing business requirements and opportunities in a way that can evolve along with the business to become a source of genuine strategic value (Link it back to Balance Scorecard or organization goals). Business users and leadership needs to be prepared to act on insights from descriptive to predictive analytic models to see true value of investments made.

About the Author: Gaurav Vasu
Gaurav Vasu is Global HR Market Intelligence & Analytics lead at a leading global IT services company . He has worked with CEO office, CHRO’s and senior HR leadership to shape the business strategy, identify human capital implications and design people practices to enhance performance and productivity. He is one of the top Industry experts in the Research, Consulting and Analytics domain.

He also specializes in Growth Consulting (IT/ITES), Market Entry Strategy, Industry Analysis & Assessment (India, China & Philippines), Talent Supply Mapping, Vendor Analysis, Peer Group Benchmarking, Financial Analysis (Discounted Cash Flows, Relative Valuations, Simulation, etc.) and Wargaming.
During his 12+ years of experience, Gaurav has helped delivered consulting by growing the research and analytics value chain in companies such as HCL, Accenture, Zinnov, and Knowledge Faber & Nirvana.