From Small Data to Prescriptive Analytics or Explainable AI, we dedicated our previous articles to explain how Small Data coupled with AI can change business for all types of companies. New techniques deliver fast actionable insights and help for better decision making. Let’s now dive deeper into Prescriptive Analytics for Small Data with a case study on one of our customers. Discover how we helped our customer save 150K€/year of labor cost in one warehouse based on 3 months only of historical data.
The right number of people at the right time
Ordering something online and getting a home delivery a couple of days later is a mainstay of our modern lives. We shop on our devices and expect a quick product delivery, or we might consider another supplier next time. But we don’t suspect how challenging it can be for the delivery companies to respect such short timespans.
Some logistics companies specialize in e-commerce. Their customers include various online retailers. We work with one of these logistics companies which store their products and take care of their shipping. They require a lot of manual workforce to prepare the orders and charge the trucks before the delivery company picks them up. However, activity fluctuates from one week to the other.
This is why they hire temporary workers to get enough flexibility to adapt to their activity needs while controlling costs. Still, there are some risks :
- if they hire too many temporary workers, they will be overstaffed and their costs will be too high;
- if they don’t have enough workforce, orders won’t be treated correctly and delivery delays will impact the service quality.
Consequently, they must anticipate activity peaks and troughs accordingly to hire the right number of people at the right time. Solving these types of issues can have a great impact on their operational efficiency. Enter MyDataModels and our Prescriptive Analytics for Small Data solutions.
Prescriptive Analytics for Small Data: gains in operational efficiency
The challenge was quite simple: help our customer build a Prescriptive model that would anticipate their workforce needs from one week to the next with the best accuracy possible.
The first step was to build a dataset with enough information. They used internal data from the last 3 months, ie. the number of temporary workers hired on a daily basis, the number of prepared orders (types of packages, of products, of orders…). They then added some external data from their customers in the same time span. The data consisted mostly of information on the number and type of orders received. By correlating external and internal data they got a comprehensive view of their activity in the last 3 months. Although, getting this view was not enough, they needed real insights.
They then decided to go further and use this data to build a Prescriptive model. The Small Data collected over the last three months enabled them to build these models. A few amounts of data, Small and Smart Data are enough to find patterns and correlations to get efficient Prescriptive models. Remember that Small Data starts with Excel spreadsheets of only 300-row and 15-column. Finally, our customer also realized that they needed to understand the recommendations of their Prescriptive models to see what impacted most of their activity from one week to another.Hence, the choice of an Explainable AI solution for Small Data such as MyDataModels’ one made sense. With ZGP embedded in a Decision Intelligence Platform, they created a Prescriptive model that was able to anticipate their activity for the next week. Compared to the previous statistical methods used, the variance of the prediction improved by more than 50%. It means that ZGP Prescriptive models were much closer to reality than the previous methods used.
Solving business issues with Prescriptive Analytics
In business terms, it means that every Thursday at 12pm our customer receives a global report on next week’s predicted activity including the number of packages and orders to prepare as well as the number of temporary workers they will need to meet the predicted activity. And since they use an Explainable AI system, they have insights on what will impact the activity and can then adapt their hirings to the evolution of their activity.
Availability was the main issue that Prescriptive Analytics for Small Data helped solve for our customer. More importantly, it had huge financial implications. For a single warehouse, thanks to our Decision Intelligence Platform, they optimized their hirings to save €150.000 per year while improving their operational efficiency. Basically, they did more with less.
But of course, Employee availability is not the only business issue that Prescriptive Analytics for Small Data can solve. Improving Operational Efficiency is an essential part of every business expert’s daily work. This is why we’ll be happy to help you uncover hidden potential thanks to our Decision Intelligence Platforms. Interested in saving money while improving ROI for your business too? Contact us now and discuss with our experts!