Why We Crave Machine Learning And AI

By: Crystal Widjaja, SVP BI & Growth, GO-JEK

Consumers have been able to express their desires better with the adoption of tech and there is no reason why businesses won’t cater to experiences that make the users’ lives better. In response to this, app and software developers are able to leverage tools that meet these expectations to present products and content that reflect a shopper’s past picks, experiences and preferences.

Artificial Intelligence backed by Machine learning capabilities and Data Analytics as services is on the rise in e-commerce in order to bring about what is termed “conversational commerce”. Whatever the platform maybe, AI implementations give shoppers a user friendly entry into their offerings. Thus the markets have also become increasingly competitive to the point that products and business models must change and adapt frequently in response to evolving market dynamics.

Why Data Analytics Matters

Whether it be an upward blimp in growth, or a downward turn, we always turn to the data when tracked KPIs change unexpectedly. Why did conversion rate suddenly go down? What city caused it? What product was most impacted? And how can I prevent this from ever happening again?

A business intelligence team at a company with a data-driven culture will cultivate, curate, and refine the data such that the business is able to drill down into the specifics of a metric and its symptoms at high granularity and dimensionality. In turn, growth signals are exploited at their fullest, and more importantly, emerging issues are triaged as quickly as data can be made available. With an extensive BI platform, we can quickly identify the root cause of declining KPIs, and ensure that we do not make the same mistake again, or put features in place to prevent it from reoccurring. As the renowned essayist and philosopher George Santayana put it, “Those who cannot remember the past are condemned to repeat it”.

Machine Learning and Data Analytics is a powerful compounded tool

Machine learning teaches digital applications and computational machines to understand patterns and predict potential outcomes from large amounts of unstructured data sets and over time, the ecosystem is able to think and resolve problems with little or no human intervention. Now a significant number of CIOs believe that smart applications is a topline priority for their businesses to not just have better control of their data but also the allows for the means to make the most of them. Algorithms are now the mature enablers of decision making in real time that wasn’t in existence a year or two ago.

Having readily available data analytics forces the organization to use data to make decisions. As a business grows and scales, the amount of data grows exponentially with it - making the job of getting insights out of raw data that much more involved. Thorough analysis of KPIs and the metrics that matter mean that the business only iterates and concentrates on features, experiments, and processes that objectively lead to more growth, rather than spreading resources so thinly across unimportant initiatives. This creates “forced prioritization” - with a limited amount of resources, development is prioritized for that which has been proven to ‘move the needle’ for the key KPIs.

Unstructured Data Explosion:  Is BI Critical for all segments?

I would say that to make sure that data is structured properly is a really important​ part of the job of a Business Intelligence unit. Having vast amounts of unstructured data is a sign that the BI team and a product development team were not coordinating early enough, and data was stored in a “just store it” manner rather than having been methodologically curated and discussed for future use. Audio, chat logs, and images, while relatively unstructured, should have some level of organizational design and cataloguing so that AI and ML can make use of it quickly with as little data preprocessing as possible.

Contrary to the popular opinion, I believe that Business Intelligence is not mission-critical for all companies - at least not the small family businesses that expect to stay small, or companies in certain fields (such as those that don’t target mass consumers and have a niche field). For everyone else, the role and vision of business intelligence should start early, start small, and be exceptionally involved in the product development from the get-go. That is, there should be a working feedback loop between BI and other parts of the company, such as finance, product owners, and marketing, and it should happen as early as possible. Because the bigger a company is, the harder it is to integrate effectively, or at all.