Pre-pandemic, companies were concerned about the ongoing shortage of data scientists. Now, with organizations freezing new hires and in some cases, shedding staff, the shortage looks to get even worse. In response, consumer product brands and marketers are trying out “self-service data science” applications.
The goal? Leverage powerful data science to maintain and grow the business, while augmenting declining internal resources. For consumer goods and retail brands considering this path, it’s important to first understand the terms and technology, how the tools function, and which applications are paving the way toward true self-service.
Breaking down self-service data science
An application that is self-service means that anyone, regardless of expertise, should be able to use it and derive benefits. Think about an order kiosk at a restaurant. Kiosks have an easy-to-use interface that hides the complex programming “under the hood.” It translates the application into language and visuals that anyone can use. It doesn’t eliminate the human at the counter, but it allows more people to access information in the way they expect it.
Data is the fuel that runs every application. This data needs to be prepped and easily accessible by the application itself. If IT or engineering teams need to manually feed data into an application, it’s not “self-service” and is not likely to achieve its desired impact. A self-service application should connect to any data source quickly and seamlessly regardless of data format, quality or size.
Science describes the complex programming and mathematics done behind the scenes that helps turn data into something useful. Within the broad category of science, automation and machine learning are more specific terms we use to describe functions of modern digital systems — intelligent systems that are able to learn and make autonomous decisions based on data behaviors, external inputs and more.
An inventory application is a good example of this. Based on data such as current shipments or orders, availability, distance, weather and more, an app can estimate delivery times and provide an update to partners or customers. As an intelligent system, it can modify and update delivery times automatically based on changes to the data inputs.
Self-service data science use cases
Self-service data science applications already exist that are available to consumer goods and retail marketing teams.
For example, demand-side ad platforms (DSPs) used to buy mobile, search and video ads from a marketplace use data integration and machine learning to optimize advertising spend on a minute-by-minute basis, learning from consumer behavior and past performance. Just specify a target and a budget and the platform does all the work. It handles the mathematics, optimization and learning, while offering an easy-to-use interface for the advertising or marketing manager. Because this exists, brands can use their data scientists for more complex customer segmentation and insights, allowing a self-service application to run ad campaigns.
Now, analytics and AI are making self-service data science tools and applications available for other consumer goods and retail use cases, alleviating the pressure of staffing shortages and helping businesses survive (even thrive) through the pandemic.
Applying self-service in the real world
One of the most beneficial applications of self-service data science is the ability for non-analyst teams to find unexpected, important and actionable insights hiding within their data. This was once limited to data scientists, but self-service has changed that.
For example, the marketing team at a large consumer e-commerce brand observed a significant drop in website page views from a particular search engine. In fact, traffic was 49% lower than the model expected and had the potential to damage sales. Normally, they would ask a data scientist for assistance in digging into the root cause, a process that would take a week or more. Instead, using our self-service data analysis platform, the marketing team was able to discover the reason behind the downward slide on their own, and quickly focus efforts to combat the trend in just one day.
The platform’s root cause analysis feature displayed possible causes for the unexpected page view drops. With this AI-driven information, the marketing team was able to recognize the paths that were dropping and identify the device category and regions.
The root cause feature also showed that desktops were the primary source of the issue, Florida was tagged as the primary region and “organic search” was a primary contributor to the downward trend. With this information automatically provided to them by the platform, they were able to modify the specific page paths driving the downward trend immediately, successfully moving traffic and revenue back up to expected ranges.
The power of a virtual data scientist
Also referred to as automated business analysis platforms, self-service tools are helpful because they perform analysis automatically, uncover data patterns and unexpected events and elevate concerns or opportunities to managers very quickly. It’s like having a virtual data scientist in the system that works to uncover trends and behaviors every day that marketing, supply chain and operational teams couldn’t uncover on their own, without intelligent AI tools.
And unlike most complex business intelligence platforms, brands can deploy self-service tools very quickly, often up and running in a few days. By simply integrating key data from systems and applications that already exist within the business, companies have an immediate foundation for better, richer insights. Even with a shortage of data scientists, consumer goods and retail brands have an opportunity to use data for positive momentum during and after the pandemic.
Sean Byrnes is cofounder and CEO of Outlier.