During one discussion I had with a person who holds. top management position in a retail business, I could hear the pain he had with his supply chain optimization. In fact, we were talking about rethinking his Information system. And as we moved through the process, he shared with me how he’s having an issue with all those inventory of unsold items that he has. In fact, not only it takes time to source the items and have them delivered to his warehouse, but more often than not, they end up stocking items that are unsold. This is a huge issue for his business as unsold items cost him a lot.
The A-Ha Moment
That discussion sparkled a lot of thinking from my side. Indeed, this must be a very common issue in the retail business. Since the retail business model highly relies on a very quick rotation of the inventory so as to make some nice results, any inventory that is not rotating is considered as money that is blocked. While sales people can do their best so as to promote the best selling items, there will always be a certain set of items that won’t sell as quickly. But how do you predict what items won’t have the expected success? What if the people who had a certain knowledge about what’s successful leave the company, thereby taking with them that competitive advantage. Worse, what if those key people goes to competition?
In fact, the top manager I dealt with was aware that he needed to setup another solution, another system, that can become an asset he can rely on. It would be a plus if that asset doesn’t rely too much on people. In fact, while he wanted to pump, so to speak, the key people’s knowledge into an information system, he had to be pragmatic: that knowledge is evolving all the time, and most of the time, asking people to document the information system will take some resource from the “productive time” of their key people, and will eventually demotivate those key people.
The options as well as the obstacles
While surfing on the internet and doing his benchmarking, that top manager knew that there must be some solution in the Artificial Intelligence field. That solutions should exist for helping him with prediction for his supply chain management. But few obstacles went on the way, for him to go further:
- the AI-based solutions came with pushy sales people that want to sell high-end solutions that may not fit with his issue. In fact, he first wanted to test-drive before deciding on the solution,
- while test-driving, he didn’t want to share his decade-long customer data history with the first AI-solution provider that come in. Indeed, that data are part of his competitive advantage asset
So how should the top manager approach the issue, and thereby give a test-drive without having to reveal his own data? And how can he “play” with the solutions without putting a huge investment upfront?
How to play with it without huge upfront investment
Big names are usualy mentioned when it comes to Supply Chain Predictive Analysis software that comes with some Artificial or Business Intelligence. His top-of-mind options were SAP Supply Chain Analytics Software , PeopleSoft Supply Chain analytics by Oracle, JDA Supply Chain Analytics. But he was afraid of the expected upfront investment. In fact, he didn’t want to deal with the back-and-forth discussions that would imply if he requested some information from those big names.
Being a bit techie, the top manager considered freemium or open source solutions for implementing a test drive. A few searches on the internet helped him identify some soutions that seemed to fit his needs. While the freemium option could be interesting, he was afraid that he won’t event have the time to get a glimpse of all the features and the trial period will be over. Moreover, since he’s not well-versed into all the techniques, he would have to go through some learning curve before being able to really appreciate the value provided by whatever solution he’ll be trying.
As a last resort, he considered going the open source way. Being a bit techie, he knew that going the open source way usually means that you are give the ingredients, and it’s up to you to make the neede assortment so as to get a nice food, so-to-speak. So he had to be psychologiclly prepared to doing a lot of research. As he moved along, he found out that H2O retail use case resonates with him. The logos that are listed under he use case gave him trust too. The issue he had was that he had no internal human resources that could work on the topic with him. He had to work it by himself.
Where do I find dataset
After rolling his sleeves, the top manager decided to go with H2O solution. While he was scroling through the documentations trying to understand where to start, one question still remained unanswered. Supposing he could find out how to make it work, how can he test without revealing his own company’s data? After all, he has no clue what would be done with his datas that will be manipulated by that open source project.
Here again, search on the internet got him covered. In fact, he found that some websites provide some open data sets that he can play with. In fact, on a website called Kaggle, he could find actual, but anonymized, dataset from actual retailers. The datas are distributed under the Creative Commons license, meaning he could use it without major licensing issues.
Is this the best solution?
The top manager I talked with have found an AI-based Supply Chain Management system. Moreover, he could find a data set to play with. He is fully aware that on of the key success factor to his project is his ability to put everything together. And this is where some specialisez help may come handy. In fact, on of the biggest challenges for implementing such project is the ability to implement all the specific need of the customer. While on this case, the open source solution is technically feasible, some people may not be willing to roll their sleeves and get their hand wet with the intricacies of setting up the system. But, even with paid solutions, one can’t ignore that integration step. Would it be a solution to specialise an in-house resource who would lead those kind of transformative technology? I would say yes. After all, that system addresses perfectly the initial issue: it transfers the majority of the knowledge from key people to the AI-based tool. Which means that the competitive advantage asset moves from people (who can go away), to a system that the company owns. And that’s a huge transformation. And I am not even talking about the awesome features that the machine learning powered solution is bringing to the table. So yes, the open source way may not be the best solution for everybody, but it’s a nice introduction path to AI. And it addresses one of the big issues retail companies face: supply chain analytics and forecast are addressed in a more structured manner.
Growing, sharing in the digital era 