Confusion about what exactly big data, analytics and optimization mean, is not uncommon. Many companies do not fully grasp what these things are and sometimes it feels as if new terminology is being used for old concepts. But even if you understand these concepts, that’s only half the battle. The key to success is knowing how to apply these assets usefully to your business, or at least know whether you should use them in your business at all.
Big terms, simple explanations
Big data simply means you have a lot of data. That’s it. There is nothing that specifies what ‘a lot’ is, and it is usually compiled from different systems like ERP, MES, etc.
People tend to overestimate the value of big data and at the same time underestimate its potential. Big data on its own does nothing for you. Analytics is the filtering, compiling, analyzing, leveraging, or otherwise using of any data – including big data – to gain understanding.
Analytics can be broken down into three levels:
1. Descriptive analytics
Basically, this is looking back at performance, examining data to gain understanding about your business. It is typically used to report on past performance. Descriptive analytics has been around for decades; the only difference is that with big data there is more to look at and the methods and ease of use have improved over the years. The problem with reporting the past is when you are expected to interpret what it may mean for the future. To support you in that process, you have to move on to the next level; predictive analytics.
2. Predictive analytics
Learning from the past is good, but the ability to forecast the future is even better. That is what predictive analytics does, looking at past patterns to predict future outcomes. For example, how demand will change during certain seasons, how different pricing strategies affect will affect customer behavior, or how long it will take a machine to perform a certain operation. However, in most cases descriptive and predictive analytics combined are still not sufficient to steer your operations. E.g. predicting that demand will change is great, but will this information not tell how to run your business. For that, you need prescriptive analytics.
3. Prescriptive analytics
So, you have analyzed your (big) data and predicted the future. Now the key to becoming more profitable is to decide how to use your resources given different future scenarios and probabilities. For each scenario, you need the best answer to questions like: What machine capacity do I need and how many shifts? In what sequence should I manufacture and in which facilities should I produce what? Prescriptive analytics, also known as supply chain analytics or supply chain planning and optimization, enables you to calculate the best course of action for any number of future scenarios. In other words, it helps you make decisions that steer your business.
Each level requires different technology. For descriptive analytics, you need a way to go through masses of data, find correlations and generate reports. Predictive analytics requires statistical algorithms – trend analysis and so on – to use what happened in the past to predict what will happen in the future.
Prescriptive analytics has to select the best solution from an almost infinite number of possibilities, which can only be accomplished with advanced mathematical optimization techniques.
Which is the right one for you?
This is, quite literally, the million-dollar question. Choosing the right technology involves a large investment and considerably larger returns.
The million-dollar answer depends on your type of business. Retail, for example, relies heavily on predictive analytics because demand patterns are extremely important. On the other hand, if you are an internet startup developing something completely new, there is no past company data to analyze.
Some markets are also less complex than others. Retail is very complex because you are dealing with millions, maybe billions, of shoppers. Whereas if you corner the market selling specialized multimillion-dollar machines, your market is better defined – you have a limited number of customers who need and can afford your product.
At the same time, simple demand patterns do not mean simple processes. If you are building big machines, the number of components and steps needed to make a single product makes production extremely convoluted. Adding another layer to the complexity is the likely fact that you have to tailor each machine to the needs of individual customers. In such cases, supply chain planning and optimization – prescriptive analytics – is vital.
As I’m sure it is becoming clear, large businesses with complex planning puzzles need to ask a different question. Rather than “Which level of analytics can add value to my business,” you may already see that integrating all three levels will give you a tremendous competitive advantage.
I’ll address that in my next blog.