Supernova Award Category
Data to Decisions
The Problem
As a non-profit, Kiva doesn’t have endless resources to spend on technology. Still, between its base of nearly four million borrowers and lenders, it manages a lot of data. Eight years after its founding, Kiva hit an inflection point upon realizing they were being inundated with new data points and were having difficulty capturing them all. This level of digital maturity required a reassessment of their internal systems and digital strategies. In 2013, Kiva officially outgrew its MySQL database and set out to research the right cloud tools to help capture data, cut down on processing time, and lighten workloads for its small but mighty tech team.
Prior to Snowflake, Kiva was unable to accurately and quickly process data because it lacked the right data warehouse. It also lacked the tools to allow non-technical members of the org to easily analyze data. Access and analysis was limited to a few individuals. Provisioning access to data was a labyrinth task of managing permissions. Since using Snowflake, Kiva has democratized access across to org pull variable and reliable data for quick, effective analysis while ensuring controls around privacy and security of data.
As Kiva struggled making decisions on where to invest efforts, it became clear that it lacked the ability to analyze data. It was unable to fully quantify efforts in email and marketing campaigns. Further, little analysis could happen because of the difficulty of easily accessing data
The Solution
To help speed up and simplify data processing, Kiva selected Snowflake’s cloud-native data warehouse to resolve the challenges it faced in the following ways:
Kiva tried a number of technology offerings before landing on Snowflake, which was by far the easiest, fastest and most cost-effective cloud data warehouse available. When evaluating data warehouses it was clear that Kiva had to find the most cost-effective solution because donors place great scrutiny on Kiva to ensure donations are spent on program costs rather than on vendor services. Snowflake was the clear winner in cost-effectiveness. Further, their excellent staff were devoted to seeing Kiva succeed and provided the help to make the organization successful.
Kiva hired a dedicated analytics manager to help maintain and improve systems around Snowflake, transforming Kiva’s entire organization into a data-driven workforce. Today, Kiva can enjoy a host of new benefits around data management.
The results
Kiva changed its entire approach to managing data with Snowflake and was able to accomplish the following:
Institutionalized data sharing – Enables consistent and reliable data reporting across 400 Trustees and Field Partners to better source entrepreneurs, screen borrowers, post loan requests, disburse loans, and collect payments.
Democratized research – Kiva meets demands from academic institutions for access to anonymized Kiva data for economic research through APIs.
Widened audience and Increased funding – Kiva uses data from Snowflake to determine the new markets to focus advertising efforts, greatly improving media purchasing decisions. Snowflake also helps Kiva maintain critical grants from Mastercard Foundation and Capital One, among others, who require detail ROI in funding proposals.
Improved lending strategy – Snowflake processes Kiva's social proof mechanism, which helps vet borrowers more effectively by understanding their online network. Once this data is warehoused, it can be shared and eventually automated to help scale analysis and serve more users.
Ensured operational efficiency and cost savings – Kiva uses Snowflake to ensure its technology is operating as efficiently as possible in the cloud so that resources go to its core lending activities instead of managing the complexities of modern IT infrastructure.
Metrics
Ensured operational efficiency and cost savings - Kiva lowered its overall engineering budget by 20-25 percent over the course of a year and a half and saw a significant saving of around 50 percent, compared to running an in-house data warehouse. Time spent on maintaining data warehouse systems dropped from a full-time job to a small portion of the operations team’s time.
The Technology
Kiva uses Snowflake, the only data warehouse built for the cloud. With up to 900 data points per loan, Kiva manages as much data as traditional banks but with far fewer resources, so it needed a strong data warehouse to store and secure its data.
Disruptive Factor
The biggest challenge in implementing Snowflake was committing to it as an organization. With new tools, it’s difficult to be effective right away and previous tools offered very little in the way of support. Fortunately, Snowflake staff guided Kiva through adoption. After that, Kiva could quickly migrate to Snowflake and begin enabling access across Kiva.
[re: impact] Kiva uses Snowflake’s data warehouse to leverage data and make better informed decisions. The impact has been substantial. For instance, using Snowflake’s data processing capabilities, Kiva developed its World Refugee Fund 2018 Impact Report, which revealed that refugees have a repayment rate of 96.6% compared to 96.8% for all non-refugee loans. The data used in this report has allowed Kiva to prove to banks that refugees are low-risk candidates for loans.
[re: what sets Kiva apart] The unbanked population is a worldwide problem, with 2 billion people globally who are considered underbanked. Business leaders are becoming more conscious of the impact they can have on these issues, and Kiva is at the forefront of that movement, especially when it comes to data sharing. Kiva uses data from its 4 million lenders and borrowers to test what works and share results to change the definition of “too risky” or “unproven” – a work ethic that requires keeping up with a constantly changing technology industry.
Shining Moment
Forbes profiled Kiva’s innovative social proofing mechanism, which helps maintain its 97% repayment rate. One of the best ways to support people without a credit history is to use social networks to fund the first 10 percent of their loan, providing an alternative way to predict the best borrowers. This entire process flows through Snowflake; Kiva is in the early stages of automating this process further with machine learning to encourage even better results.