“We Have Data, But We Don't Know How To Use It!"

Justin Chua
Justin Chua

Canva - Man Holding His Head in front of Laptop

Everyone is talking about Artificial Intelligence (AI) these days, but few enterprises really understand how they can actively adopt it. Gain a head start here. 

Quick Links
#1 Descriptive Analytics
#2 Predictive Analytics 
#3 Prescriptive Analytics 
Great, but where do we begin?

"What can I do with my data? Is AI even a possibility?"  

Does this sound painfully familiar? If yes, you're not alone. Many companies often ask us this during the early stages artificial intelligence (AI) discussions.  

It is surprisingly common even in companies with established data science teams. Data talents are in high demand, scarce and, unfortunately, expensive. With limited resources, businesses need to prioritize key business challenges to answer in order to effectively determine the type of analytics needed, and if AI is even required at that point in time. 

In this post, we share how the different kinds of analytics can be used on your data to obtain the answers you need. 

#1 Descriptive analytics

“You can have data without information, but you cannot have information without data."
Daniel Keys Moran, author

Descriptive analytics, as its name suggests, describes your business' existing data.  

This technique is most widely used in today's businesses. For some, it’s the only analytics used. It can be as simple as your staff generating an Excel report and describing its content to you. However, we often hear complaints of inconsistent reporting and how reports get lost or corrupted.

Asides from the regular report sent out via email, an effective approach to descriptive analytics is via a collaborative centralized Business Intelligence (BI) dashboard where reports can be automated and easier understood visually. The process of semi-automated data processing reduces human-errors and can act as a single-source-of-truth for everyone, including the leadership team. 

Companies that have advanced themselves have gone a step further by applying AI to identify hidden relationships within vast amounts of data. 

For example, a client of ours launched a new product across 10,000+ stores in Indonesia. While the launch was a huge success, the company needed to track its performance across the country and to sustain its sales growth better. Questions like 'Which stores have marginal profitability but are showing high-growth potential?' were difficult to answer. Hence, we went on to build an analytics model that laid their AI foundations all delivered through a customized BI dashboard. Through the model and dashboarding, we managed to profile all 10,000+ stores into different group types (with Unsupervised Machine Learning) and presented them to the clients in an understandable way. Adopting the concept of the 'dog-star matrix' we were able to identify stores as 'Cash Cows', 'Rising Stars' and 'Problem Childs', all wrapped together with a geo-location heat map, which allowed leaders to better strategize business plans, marketing efforts and inventory. 

#2 Predictive analytics

“The goal is to turn data into information, and information into insight.”
– Carly Fiorina, former president of Hewlett-Packard Co

Historical data can tell us some of the future. By identifying recurring patterns within your data, our AI models can predict key outcomes for you.  

Today, predictive analytics is cleverly applied by leading-edge companies across a wide variety of use cases to stay ahead.  

For example, dating apps are using trends of similar profile to predict how likely a person is to date another. Cutting-edge manufacturers are spotting machine failure before it occurs and servicing them in advance to keep production uninterrupted. Savvy supermarkets are forecasting product demand to manage stock inventory and cash flow. 

One consumer goods company we spoke to was concerned with the supply of its raw materials. While sea freight is most cost-effective, timeliness was an issue. If shipments of raw materials were delayed, production would be affected, and its store faces lost revenues due to insufficient stock. By processing key data points, we were able to predict estimated arrivals and potential impact to distribution and sales with AI. This approach allows the supply chain to mitigate risks by activating contingency plans early. 

#3 Prescriptive analytics

“If we have data, let’s look at data. If all we have are opinions, let’s go with mine.”
– Jim Barksdale, former Netscape CEO

How effective would you be if you always make the best decisions? And how sure are those decisions? What if you can automate parts of your decision-making process and stay focused on more complex issues at work? 

Prescriptive analytics is the next level of decision-making. It calculates the impact of possible actions and recommends the best option to achieve desired outcomes.  

Google, Amazon & Facebook are famous for scaling their businesses with prescriptive analytics. Facebook's content recommendation keeps users engaged for longer while Amazon’s account for 35% of their revenues at one time. Google's search engine processes 3.5 billion searches every day and prescribes the relevant results to each user. 

How we applied similar logic can be seen in this example. One of our engineering-based clients wanted to reduce maintenance time by providing on-ground service teams with a prescribed list of actions in order to optimize resources. This lessened the time engineers spent searching for problems and solutions and enabled them to prioritize work effectively. By predicting faults likely to occur, our client could further improve their effectiveness by addressing unseen core issues ahead of time, looking into areas where the human eye may miss.

Great, but where do we begin?

Start your AI adoption by first focusing on key business problems that impact the long-term. Create a priority list of questions and identify which of these have sufficient data to help answer them. Not sure about the types of data needed to answer your questions? No worries, we consult and architect AI frameworks that can be implemented from foundational levels. 

Once done, work with a data science team to develop a Proof-of-Concept (PoC) to test its impact within a time period. Positive results from the PoC can then help you secure internal support for wider scale adoption. 

If you need support in ideating or developing a PoC, drop us a note. DataVLT helps companies build AI solutions to address business challenges.  




The Forefront of Data Analytics — AI Builders of the Future 
DataVLT provides artificial intelligence as an outsource solution. We exist to help enterprises accelerate their growth through digital transformation by way of data analytics. 

Learn more at www.datavlt.com. 


Sign Up For Our Newsletter