Preparing your business for Machine Learning
A couple of years ago, I was speaking with a CTO of a pre-product company that was looking to hire an analytics lead. His team of engineers was working on the product, which was to be launched in 6 months. He had no user data available for the analyst to analyze, yet he wanted an analyst in his team. The analyst's role at the time was to work with the engineers to ensure incoming data will be collected, integrated, and ready in the data warehouse as soon as the data starts flowing in. This was all the analyst was to do before starting any analysis. I was pleasantly surprised at his initiative and I commented, "Someone must have got burnt before". He laughed and said, "It is I".
The truth is, it takes more than just data to be ready to be data-driven, integrate predictive analytics, or machine learning. It needs a change in mindset in the organization, starting from engineering to customer success, from an individual contributor to managers.
In this post, I will share my thoughts on best practices towards a data-driven business.
Purpose:Leverage data every day across all business functions
Guiding Principles:
I. Treat data as an asset and have a single point of ownership.
II. Ensure seamless integration, data governance, and user privacy.
III. Define business metrics and set KPIs to measure business goals.
IV. Build a consistent, scalable reporting infrastructure to benchmark performance.
V. Be agile to deliver iterative value to the business frequently.
VI. Empower teams with data skills towards data-driven decisions.
VII. Integrate predictive/prescriptive analytics in business decisions.
I. Treat data as an asset and have a single point of ownership.Owner: CDO/ Head of BI
The data owner, either the Chief Data Officer or Head of Business Intelligence, must have intimate knowledge of all transaction and analytical databases/ architecture and collaborate with business stakeholders to drive decisions from data.
Transactional Databases:
Data warehousing and Analytics:
II. Ensure seamless integration, data governance, and user privacy.
One of the foundational requirements of data-driven decisions is seamless data integration. While organizations are embracing digital transformation, data governance is becoming a challenging endeavor. Data teams face a difficult task to make data available to business stakeholders, analysts, and data scientists while ensuring compliance with external privacy regulations, industry standards, internal data using best practices, and most importantly, respecting user privacy.
Effective data governance ensures that data is consistent, trustworthy, and respects user privacy. Achieve a data governance strategy that delivers trusted information on time through the following initiatives:
III. Define business metrics and set KPIs to measure business goals.
A business metric is a quantifiable measure that businesses use to track, monitor, and assess business processes' effectiveness. A key performance indicator (KPI) is a business metric that evaluates and measures the performance of the Key Results of an OKR(Objectives and Key Results). Often, a KPI that needs improvement will be a starting point for creating an OKR, and it will become a Key Result of an Objective. KPIs are intended to measure the health of the business initiative.
IV. Build consistent, scalable reporting infrastructure to benchmark performance
An effective business dashboard is a single source for the truth to understand the state of the business. To implement a comprehensive business intelligence platform, organizations must design an effective reporting and analytics framework. When done right, a well-designed dashboard can help align organizational efforts, make fast, data-driven decisions, improve business performance, and increase ROI.
VI. Be agile to deliver iterative value to the business frequently.
Once a company enters its growth stage, stakeholders will be continually in need of data and analysis. Incremental delivery of reporting and analysis projects following agile methodology may help the congestion for the data team. With each iteration, both data and business teams benefit from incremental learning. The data team then refines and adds new signals for the next incremental improvement.
VI. Empower teams with data skills towards data-driven decisions.
With the explosion of big data, it will be challenging to rely only on the BI team to provide data and analysis. In many successful data-driven companies, every other business user can write SQL and perform data wrangling. It expedites incorporating data-driven decision culture in the organization. Today, knowing SQL is an advantage. In the near future, not knowing SQL will be considered a disadvantage. Invest in continuous learning for all data and business professionals alike.
VII. Integrate Predictive/Prescriptive analytics in business decisions through machine learning.
Once a consistent, scalable, and automated reporting infrastructure is in place, prepare your analytics team for prescriptive/predictive analytics. A few critical indicators of finding success in predictive analytics are: