Sigmoidal Case Study

ML & Predictive Analytics for Supplier Selection

Read how we helped our client halve the time spent on research and supplier selection. We had leveraged a hybrid approach to reduce risk - the most complicated cases were automatically flagged for manual verification.

Background

Free Up Your Time

With both a great number of suppliers and a high number of variables, it can be quite overwhelming for a human to make such an analysis and decision, and often some vital information might be missed. Selecting the supplier and maintaining a proper relationship can be difficult.

We leveraged machine learning to reduce the time spent on research and risk assessment from 5 hours to 3 hours a day on average.

challenge

Reduce the time needed for research & risk assessment in the supplier selection process.

SOLUTION

A hybrid system with the majority of cases being handled automatically. More complicated cases were flagged for manual verification.

Value Delivered

From 5 hours to 3 hours

On average, for an individual, the time required for research and risk assessment was cut down from 5 hours to 3 hours a day.

Greater precision

A much higher number of suppliers is automatically compared. As a result, helpful, non-obvious patterns can be found.

Informed Assessment

The solution allows our client to predict outcomes based on historical data, enabling him to make better-informed decision.

Business Problem

Our client wanted to create a platform that would allow him to compare and choose suppliers, maintain them, then continuously assess and manage risk in a more informed manner. The main business goal here was to allocate less time for research and prevent potential financial losses that could be caused by inadequate supplier selection.

A crucial part of purchasing management and enhancing supply chain sustainability is supplier selection. This can be a complicated process with any number of variables to consider. Also, it can be increasingly difficult with the growing number of available suppliers.

With both a great number of suppliers and a high number of variables, it can be quite overwhelming for a human to make such an analysis and decision, and often some vital information might be missed. Selecting the supplier and maintaining a proper relationship can be difficult.

Analysis

Risks To Consider

There is a significant amount of supplier-related risks that have to be taken into consideration. These risks types include:

Financial: Risk involved with using a third party supplier. There can always be a problem of tax changes, bankruptcy, or price change. If the supplier does not meet expectations, you face financial losses. A supplier overseas means fluctuating foreign exchange rates.
Delivery: include government or economic issues, natural disasters, etc. Such risk cannot be predicted and avoided, and they can prevent a supplier from delivering goods.
Compliance: it is essential to ensure that a company you choose is compliant with standards and regulations. Otherwise, you can face fines.
Image: it is also vital to ensure that the company that you choose is ethical and professional.
Quality: quality and performance supplier risk may involve over-promising and under-delivering, violating contracts.

Implementation

Choosing Variables

There are a lot of parameters based on which you can make a comparison. We decided to include quite many of them, so although our client named the ones that are crucial to him, he still had a chance to look at different aspects of the chosen company. Variables included:

product price – most basic variable, which everyone takes into account
tariff and taxes – if the supplier is from another country it is essential to know government regulations
product quality
percentage of faulty items
delivery performance
stable delivery
lead time
flexibility
responsiveness and customer response
commitment to quality
The financial position of the company,
Market reputation
Geographic location
The political stability of the country
Exchange rates
Environmental factors

It was essential to establish which variables are most crucial for the client. It was also essential to discuss the value of each variable with the client, as some of them would have more impact on the predictions and recommendations.

A “best scenario” is subjective. If a client’s attitude is “price over quality,” then we would focus mostly on price, but choose the best scenario considering other factors. In this case, we managed to make it flexible, choosing the most important factors for each specific scenario.

On our platform, the client can use individual variables: e.g., only the lowest prices. Sometimes, however, the best price is also classified as “risky.” For example, the best price can be offered by a supplier from a destabilized country or by a supplier with negative customer reviews.

The client can also browse through ranked recommendations of scenarios.

“With both a great number of suppliers and a high number of variables, it can be quite overwhelming for a human to make such an analysis and decision, and often some vital information might be missed.”

Solution

The big picture here was to leverage machine learning to make the supplier selection process more predictable and intelligible. We created a platform providing the best scenarios for supplier selection and risk management, based on parameters that our client found the most important, such as product price or lead time. The solution serves as an augmentation of the whole process with a final decision made by a human.

The system has two main components: the prediction engine and the recommendation engine.

Data sets generated from SRM actions, such as supplier assessments, audits, and credit scoring, provide an important basis for further decisions regarding a supplier.

Conclusion

Thanks to the AI-based sentiment analysis system, we augmented the process, significantly reducing the workload needed for research and analysis. Baillie Gifford can now use additional investment signals to their advantage, getting it directly from users, allowing managers to react to market behavior in a more informed manner. Ultimately, it leads to better portfolio management and risk minimization.