Today, we will discuss how to increase the return on investment on your data science projects. As we will see, increasing your returns simply comes down to asking the right questions. Let’s get started!
First, you will need to decide which type of analytics your project belongs to.
The Four Types of Analytics are descriptive, diagnostic, predictive and prescriptive.
A descriptive project is one that tells you what has happened. This is useful for operations management and for keeping track of what’s going on.
A diagnostic project is one that helps to establish the root cause of a problem. This is useful when troubleshooting issues in your organization.
A predictive project aims to predict the future based on historical data. This technique can be used to automate business decisions that are based on algorithms.
A prescriptive project is one that helps to make complex decisions. Multiple outcomes will be simulated and evaluated, so you can pick the most favorable one.
Doing this helps to choose the right team members for the job. Alan Jacobson, Chief Data Officer at Alteryx, elaborates in a podcast: “A data science team should contain a hybrid of specialists who can do prescriptive, predictive, and descriptive analytics.”
Next, we look at the factors that influence returns on data science projects. The more accurately you can calculate their impact on revenue, the easier it will be to improve your returns. This also helps to identify which opportunities are the most worthwhile to pursue. Jacobson continues: “Calculating ROI in advance can prioritize projects''.
Factors Influencing Returns:
- Cost savings: how can this project reduce expenses for my organization? Can I reduce expenses without impacting the quality of my products?
- Quality increase: can this project increase the quality of the goods and services I offer to my customers? How will that translate to higher revenue?
- Speed of operation: can this project speed up internal processes so we can work more efficiently? Which metrics will I use to measure speed of operation? How can we make sure it doesn’t impact the quality of our products?
- Security (limiting downside): can this project successfully avoid expensive security mistakes? How expensive would a failure be? How certain are we that it will be prevented?
- Automation (outsourcing): can this project automate repetitive tasks? How will this increase our revenue? Can it end up costing more if failures happen?
- Differentiation (innovating): will this project help us differentiate our products from competitors? How will this translate to a higher revenue?
If you can define more factors, that’s even better. Just make sure you can accurately estimate how it will impact your revenue!
Baxter Cochennet created a comprehensive article on how to measure ROI in AI. He distinguishes between four different types of ROI, including traditional (dollar amounts), non-traditional (positive branding, etc), hybrid and uncalculated ROI. Keep in mind that one project might have multiple types of ROI. Check out his article for more useful advice!
One last recommendation is to employ a centralized platform where you can track how much you’ve invested in data science projects and how much returns they generated. This will make it easier to spot the biggest opportunities to increase your returns.
To quote Hilary Mason: “In an excellent strategy, the projects will include automation, efficiency and performance improvements, but they will also include projects and ideas for new revenue generation and entirely new businesses driven by your unique data assets.” Check out her article to discover how to build an excellent data strategy.
In summary, to improve returns on data science projects: decide which type of analytics it belongs to. Assemble a hybrid team of specialists. Accurately define the factors that influence ROI. Try to estimate how improving each factor impacts returns. And keep everything in a centrally managed platform so you can keep track of progress as you go!