Software

ML for Demand Prediction with AWS

A SoftServe Client Interview with Ray Gabriel, Director of IT Infrastructure McCoy’s Building Supply

There are two great motivators for change: pain and desire for growth. Far too many companies are stuck in an even-keel mode not wanting to “rock the boat”, unaware of the benefits of modernization, or most often—they lack the manpower, tools, or expertise to take effective transformational action.

While many organizations take action due to business challenge pains like waste, missed opportunity, or competitive lag—sometimes a business that is already doing well simply wants to do better.

McCoy’s Building Supply is one of those companies unwilling to rest on its laurels.

Founded in 1927, McCoy’s is a fourth-generation, family-owned supplier of lumber, building supplies and farm and ranch equipment. The company provides a complete array of services to its customer base including consumers, builders, contractors, repair/remodelers, and farm and ranch folks. The San Marcos, Texas-based retailer is one of the largest family owned businesses in the building supplies industry.

We spoke with Ray Gabriel—Director of IT Infrastructure at McCoy’s—to learn the company’s motivations for adopting machine learning (ML) for demand prediction.
 

For those unfamiliar with your brand, please tell us about McCoy's, your business goals, and the challenges that ML is helping to solve?

RG: McCoy’s is a family owned building material retail business that is over 90 years old. We operate in five states: Texas, New Mexico, Oklahoma, Arkansas, and Mississippi. We have 88 retail units, 2 Door manufacturing facilities, and distribution sites.

Our goal is to use machine learning to help us control our inventory levels and cash flow more effectively. We want the ML model(s) to help us predict optimal inventory levels for our major product categories, while not tying up cash and floor space.
ML for Demand Prediction illustration

Good enough is never good enough.

McCoy's continually invests in its information technology as a differentiating factor for the business, and for its customers. As part of this effort, McCoy’s identified a potential way to use emerging ML technology to improve how the company forecasts inventory demand at store locations.

The hypothesis was that a demand prediction approach leveraging Amazon Web Service (AWS) machine learning (ML) modeling services could further optimize the process for buying inventory and having it stocked at the right store, at the right time.
 

What attracted you to AWS for this project?

RG: Our previous experience with AWS came with a POC to move our Data Warehouse to AWS. We ran into latency issues while moving large amounts of data daily. It did not make sense to make that shift. But although we did not move forward with that POC, we were still intrigued as to how AWS could help us leverage some process efficiencies.


Even best laid plans don’t always work out as intended the first time. Fortunately, the conversation and collaboration between McCoy’s and AWS continued—thanks in no small part to Amazon’s success in using its own AI/ML and solutions for serving its global customers as intelligently and cost effectively as possible.
 

What made you decide on AWS machine learning services—like Amazon Forecast—to improve your retail solution and achieve your goals?

RG: It is evident that Amazon does a great job in managing their own inventory; so, we assume their success is being leveraged by their own ML services and forecasting tools. With this in mind, it makes sense to take advantage of the data structure and modeling that Amazon has proven out and will continue to improve on.


As an Austin, Texas-based, AWS Premier Consulting Partner with numerous competencies and an AWS Ambassador on it global team, SoftServe’s ML/Data Science team was brought in to deliver strategy and implementation of the project.

By leveraging an AWS demand prediction model, McCoy’s can now better:

  • Free up previously locked capital
  • Reduce excess inventory
  • Lessen product shrink

…and can reduce out of stock scenarios to improve sales performance.
 

How long have you been working with SoftServe and what has been your experience?

RG: We have been working with SoftServe for well over a year. We appreciate their approach on providing solutions for business-critical operations. Their step by step process, ability to collaborate with us, and their external knowledge sources is what led us to partner with them.


SoftServe’s Intelligent Enterprise COE teams, including experts in Big Data & Analytics, Experience Design, Cloud, Business Analysis, and AI/ML—as well as an APN Ambassador—collaboratively enable companies to become truly data driven, and not just data aware.

In addition to predictive forecasting models, these cross-matrix teams deliver transformational solutions like Actionable Insights Ecosystems, Enterprise Data Platform, Big Data Cloud Migration, and Intelligent Automation—powered by AWS.
 

Before we go, is there anything else you would like to share about SoftServe’s performance with solution design and implementation?

RG: SoftServe listened to our needs and took into account our current capacity and how we would like to utilize our data more effectively. They provided more than one approach to the challenge, while giving candid opinions on each solution. We would definitely recommend SoftServe as an AWS Consulting Partner to others …without giving away our secrets, of course.
WHAT
does “data driven” mean to you?
WHO do you currently have working on machine
learning advancements?
HOW could predictive forecasting benefit your
company?

Let’s talk about your machine learning journey and how AWS and SoftServe can deliver predictive, actionable insights for improved efficiency and profitability at scale.

PDF Downloaden