EyeOn: Accelerating Business Operations in an AI-Driven World

Forecasting and planning are some of the very oldest use cases of modern statistics — businesses as far back as the 1950s used computer-based modeling to anticipate risks and make decisions. But in the age of AI and algorithms, older modeling techniques fail to incorporate the wide variety of data sources needed to produce results precise enough for the modern enterprise.

Learn more: The BI to AI Shift has Happened: How to Catch Up & Why it Matters

In addition, traditional forecasting and planning methods can be wrought with manual processes and, therefore, unintended bias. In order to be more exact, these manual processes and decisions need to be removed entirely to make way for truly data-driven decisions.

EyeOn is a management consulting company specialized in integrated business planning, supply planning, demand planning, and financial planning with more than 150 customers, including KraftHeinz, Philips, Stryker or Cargill. The company has shifted their methods surrounding data processes to keep up with today’s increasingly competitive and AI-driven world.

EyeOn’s Challenges 

For the data science and solutions team at EyeOn, which is made up of 16 people, one of their biggest challenges is the quality of the data they receive. They often have very few data points on which to base predictions, and multiple data sources that must be joined together, which means that they have to do a lot of data wrangling in order to even get to a good starting point from which to build models.

Read more: Too big for Excel? An alternative for analysis

In the past, the team had to employ a veritable cornucopia of tools to serve their clients and deliver planning and forecasting reports at a regular cadence. The amount of different tools, developed by various people over the years proved to be unwieldy when dealing with complex datasets. In addition, processing power (working on personal laptops was common), security, and data integrity were major hurdles.

EyeOn decided to make a change not only because of these challenges, but because they wanted their business to be more scalable, agile, and innovative. Given the advancements in the planning and forecasting space, having the flexibility to pivot and use new technologies and data science techniques represents a huge advantage.

Emphasising Data Quality and Business Knowledge

With lots of companies transitioning from legacy to new data storage systems, data engineering challenges abound. For EyeOn, improving data quality overall is a key goal that is vital not only to their team’s ability to deliver accurate forecasting for the customer, but for other core aspects of the customer’s operations as well.

The data team is, therefore, made up of people of various skills like statistics, engineering, or machine learning, but the underlying theme is business expertise as well. On any given project, EyeOn assigns a mix of profiles depending on the needs of the business and the nature of the project.

What unites the team is that we want to use data to make better decisions. It doesn’t stop with just making correct prediction; it stops when we use those predictions to drive a better decision — reducing working capital, increasing customer performance, reducing manual effort, etc.

Michiel Jansen,
Supply Chain & Data Science Lead| EyeOn

How EyeOn Uses Dataiku: 

• Building standard project templates
• Documentation and data governance
• Data blending and manipulation
• Data quality verification and automation
• Plugins for custom recipes and processors
• Code and visual recipes

Read more: Dataiku for data preparation and enrichment

One particular example of their newfound ability to scale and innovate is an inventory project, for which EyeOn uses Dataiku. They collected nine different datasets (e.g. inventory data, sales data, shipment data, forecast data, etc.), each distributed over one or more files per plant for 67 plants — in total, the team loaded more than 600 files. Dataiku not only provides the interface for quickly loading all files into one dataset, but it also immediately identifies files that aren’t in the right format.

Read more: Dataiku for data preparation and enrichment

Since Using Dataiku for More Than a Year as Their Centralized Tool of Reference, the Team at EyeOn has Been Able to:

• Shorten the time it takes to onboard new team members.
• Develop a modular system of building blocks and tools that they can use in different ways depending on the customer and the question they’re trying to answer.
• Sure up data governance policies and practices to ensure everyone is working in the same way with the same processes, including documentation and dataset tagging.
• Build different standard data project templates (for forecasting, inventory, etc.), which significantly shortens the time-to-delivery for customers.