Modern Software Asset Management: Predict Your SaaS Needs and Avoid Overspending

Less than half of organizations have methods to track and rightsize their SaaS usage. The same percentage of organizations offboard employees from SaaS apps they don’t actually use. What’s the result? Nearly one-third of SaaS licenses purchased by organizations go underutilized or wasted, according to the State of ITAM Report by Flexera.

On the flip side, some organizations have implemented software asset management (SAM) solutions to save money on their SaaS purchases. For example, organizations with up to 20,000 people have saved an average of $2 million on SaaS per year. Snowflake itself has saved an estimated $5 million, and we employ just over 5,800 employees (as of January 31, 2023).

So how do we do it? Like most organizations, we use dozens of SaaS applications that span every department and nearly every use case. (Larger organizations regularly deploy hundreds of SaaS apps used by thousands or tens of thousands of employees.) In turn, efficiently managing our SaaS usage was not possible and purchasing licenses per the SaaS vendor’s model became an outdated practice. 

To solve this problem, we needed a way to monitor SaaS application license consumption, and identify ways to get value for our money, while serving the needs of our employees,  our customers, and our business. With the help of one of our in-house data science teams, we built SnowPatrol—the next evolution in SAM solutions.

A machine learning (ML)-powered application, SnowPatrol is an end-to-end automated SaaS license lifecycle management tool optimized for cost. SnowPatrol combines data from sources such as Okta, Docusign CLM, ServiceNow, and Workday. Using this data with an ML model, SnowPatrol can predict if an employee is likely to use an application in the next 30 days. If not, we revoke that license. SnowPatrol handles efficiencies and gives us insight into who is using which applications and how often. All aspects of the user experience are monitored and fed back into the model for sustained optimization. Armed with these insights, we can monitor employee experiences proactively, and predictively provision applications for both existing and new employees.

For a CIO, having visibility into SaaS portfolios is a crucial step toward application portfolio rationalization. SnowPatrol enables our IT teams to partner with Procurement on every new SaaS app we intend to buy and every app we consider for renewal. This framework to optimize and rightsize helps with forecasting and translates into cost savings.

SnowPatrol is a truly native Snowflake application. With data ingested and processed on Snowflake, the model is trained and built using Snowpark, our developer environment, and has a front end based on Streamlit, our open-source app development framework in Python. All of this was built, deployed, and monitored by a team of just four Snowflake employees. 

In the year since SnowPatrol launched, we achieved 60% reduction in tickets, 50% improvement in employee experience, and $5.5M in SaaS cost avoidance. The inherent capabilities of Snowflake as a data platform allowed us to unify siloed data from multiple apps and use this single source of truth to provide insights and value previously unobtainable for us. ML application development is notoriously difficult, with a majority of time spent reiterating costly models and monitoring them. But with capabilities such as Snowpark, we accelerated our data processing and leveraged it for a greater velocity in prototyping and experimenting.

Building the SnowPatrol front end to tie in the entire framework via a Streamlit application enabled us to empower everyone with actionable data and insights, and make generally complex solutions more tangible and intuitive. SnowPatrol provides cohort analysis at a role or department modality to better understand usage, which in turn facilitates automated access provisioning. Using the front end, organizations can scrutinize the model performance parameters and tweak them to their specific demands. The ability to forecast for renewals based on the true demand of the organization delivers real savings, making SnowPatrol the next evolution in SAM tools. It essentially converts the traditional SaaS subscription model to a consumption-based one.

Our next steps are to leverage our learnings and provide them as a native application so that  all our customers can achieve what we did: rationalize their SaaS application portfolio and dramatically improve their employee experience.