I interviewed Carlos Panitz who discussed Analytics Applied to Demand Management.
It's nice to speak with you again, Carlos. This is our next topic in the series that we're doing with you. The topic is analytics applied to demand management. My first question is, can you talk about why have analytics been a buzzword over the last few years?
I think that, in the first stage of supply chain management, most companies and also software providers try to address the classical problems of supply chain regarding demand management, providing frameworks for make-to-stock companies with forecasting tools, trying to identify patterns in the past and then extrapolate for the future, and then protection planning and material planning. This is usually called the supply chain suites on the S&OP and supply chain softwares available.
But after that, everybody starts to ask, what is the next question once a supply chain is not a single entity? It's a group of companies from suppliers to distribution channels that you need to coordinate and make all these entities work more synchronized to meet demands. And then, the answer is these original, classic frameworks of forecasting, protection planning, and material planning, it's a very good start but do not support all the dynamics of the market regarding the change in the consumer behavior and the changes in the competition environment.
So a few years, since maybe in the '80s, many small companies and universities started to develop and bring some algorithms and some mathematical models to address some practical problems on the demand management. Improving forecasting approach, trying to improve the replenishment process with the VMI, with the collaborative planning forecast and replenishment framework and, more recently, using some algorithms of machine learning, narrow networks, another approach, to try to refine and improve the capacity to understand demand and the way to meet these demands.
The other aspect is the internet and the cost to exchange information goes down very strongly in the last few years, which allows to deal with more information with a much lower cost than in the past. So this enabled some business case to exchange information, to share information, regarding the inventory, visibility, to sell out information and try to use some tools. Get more data, get more insights from the same data available on the database.
How can this approach add to the current frameworks that are currently available?
Providing insights. So the answer for this question, I would summarize on that word --provide insights. Provide the capability for softwares to not only create [inaudible 00:04:12] reports but also create warnings, alerts, and try to refine the results of these algorithms. This is pretty much the way that the analytics can improve the current framework and also try to boost the capacity of an analyst in a company to make questions and get answers very quick because... In a very user-friendly screen, it's possible today, with some analytics tools available in the market, create scenarios, cross data, and try to get insights.
Another important thing that usually came with some analytics approach is try to analyze a huge mass of data and mixing internal data of the company with public data available to try to identify patterns that, in the past, would be very costly to do by not having some frameworks and a lower cost for trying to make all this analysis in a short period of time.So this is the way that analytics can improve the current framework -- providing insights for analysts and allowing them to make whatever [inaudible 00:06:05] they want. They will be able to create the scenarios and analyze the scenarios and try to get some answers for information that usually, transactional, our businesses systems cannot provide, by definition.
Can you provide an example of analytics supplied in supply chain?
I think everybody or most people and listeners of your program have the experience to buy some product at Amazon.com, for example. You realize that the more you buy on their website, the more the offers will be for things that you might like to have. So this is an example of a technique called market basket analysis, in the which you analyze a huge amount of data, try to identify patterns, and match these patterns with the preference and the history of consumption of each customers. So each customer will have a specific profile that will cross with this [inaudible 00:07:32] identifies by this kind of algorithm in order that the website, the system, or even the layout in a grocery store, have the products organized in a way that influence the demand and shape the demand based on hidden patterns that this kind of algorithm can identify.
There is, today, dozens, hundreds of examples of companies that are using this kind of technique of market basket analysis to identify patterns based on the consumption of each individual customer and then create the rules that can be applied for each of them individually. Again, this information can be refined based on public data and demographic data in order to target campaigns and promotions based on patterns. And the most important thing about that is that the raw data to use for analytics is the same raw data that was available for many, many years, and it was used only for the regular statistical process. You can get a completely different information from the same raw data based on the kind of approach that you use over that.
Thanks, Carlos, for sharing today. Did you want to also provide a brief background of yourself?
My background, I have degrees in engineering and business administration with a master degree in industrial engineering and specifically on operations research. I worked for 15 years as a supply chain manager in the [inaudible 00:09:45] industry. And in the last six years, I have been working with supply chain and S&OP software and consultancy here in South America. Thank you very much, Dustin.
About Carlos Panitz
VE3 - Supply Chain Consulting and Solutions