Funding Circle discovers key insights from large datasets in minutes

Funding Circle turned to Point Sigma, the world's first autonomous AI-driven data analytics tool that configures itself, to help them identify data-driven insights they needed ranging from marketing and credit risk through to customer service and collections.

Point Sigma is helping Funding Circle save time and discover insights to improve its understanding and decision making processes, as well as testing hypotheses around end-customer demographic, geographical and affluence data. All via self-serving, intuitive, explorative visualisations, graphs and dashboards that are set up automatically, without needing input from data experts.

Funding Circle

Funding Circle connects creditworthy businesses looking for finance with organisations with money to lend. Using a sophisticated machine learning and technology platform, Funding Circle is revolutionising how small businesses access finance.

Since its inception, Funding Circle has provided 13.7Bn worth of business loans to 120,000 businesses worldwide, and in 2021 it is estimated that lending through Funding Circle's platform supported 100,000 jobs in the UK alone.

Point Sigma

Point Sigma is the world's first fully end-to-end autonomous AI-driven data analytics solution. Point Sigma empowers people to become citizen data scientists by broadening the reach of analytics to all decision-makers, creating competitive advantage.

Customers connect their data to the platform, and, within minutes key insights are presented back via self-serving, intuitive, explorative visualisations, graphs and dashboards - all without the need for highly skilled and sought-after data experts.

Data at Funding Circle

An important reason for Funding Circle's success is their ability to access and analyse large amounts of data. Dave Clarke and Rohan Karunanayake in Funding Circle's analytics team help their colleagues get the insights they need, ranging from marketing and credit risk to customer service and collections.


Rohan Karunanayake

Head of existing borrower and investor analytics Funding Circle


Dave Clarke

Graduate Data Scientist - Analytics
Funding Circle

The analytics challenge

Funding Circle has built an impressive collection of data sources, ranging from companies' accounts, payment performance, marketing responses, credit use and more. In a rapidly changing world, with new data becoming available continuously, there is an ongoing need to connect, interpret and make sense of all the different data.

Rohan explains: “My team provides insights across the business and looks at all our data. We are always on the look out for how enhanced data and advanced analytical techniques like machine learning can improve our business performance or customer experience. Part of this exercise is to test out new data providers and see where the data can be predictive. Whilst this can be done through building hypotheses of where the data could be valuable and testing these, the approach is resource intensive and a solution which automatically tests all hypotheses with minimal analyst intervention would be very useful in speeding up this process or allowing us to test more providers per year.”



One of the datasets in Funding Circle's data universe is the vast collection of public data from government organisations, such as local and national governments, regulatory agencies and other public bodies. This data is provided through a company called Doorda, which collects, cleans, structures, documents and models this ever expanding and evolving universe of public data. “We know the value of this data, and already use it in our machine learning models. The question is; do we use all relevant insights, and are we not missing anything important that could help us make faster and better decisions?” asks Rohan.

Doorda Data at Funding Circle

Funding Circle carefully tracks its performance through a number of key indicators, such as campaign response rates, application success rate, credit scoring, payment performance and more. Rohan's team looks at past outcomes and asks; “Could we have done better if we had used more data? For example, could we improve marketing response rates or better assess credit risk by focusing on aspects of a business that we could derive from the Doorda datasets?” While the Doorda data is well- structured and documented, due to its size and variety, exploring this dataset manually to find such insights would take weeks to months of analysts' time. Moreover, a manual approach would only cover questions that the team could come up with.

What Rohan and team needed is a platform that:

  • Ingests data from different sources
  • Automatically processes, formats, cleans and combines that data
  • Identifies all relevant relationships in the data, and provides visual exploration
  • Allows users to quickly test their own hypotheses and models
  • Allow users to store, share and discuss the results

Point Sigma

Point Sigma is a data analytics and business intelligence platform that configures itself using Artificial Intelligence instead of human experts. Point Sigma uses a novel type of AI, called Artificial Curiosity™ that works


in a similar way to how humans find interesting insights in data. It works out from raw data, all the way to insights, how to combine the relevant data processing steps to produce the most interesting results.

Just load the data

After seeing a demonstration of Point Sigma, Rohan decided to use Point Sigma's analytics platform to find insights across their own performance metrics and the Doorda datasets.

Dave, an analyst in Rohan's team, created an account on Point Sigma's online platform and uploaded anonymised datasets with the key outcomes of each business across a range of metrics. Point Sigma's platform encrypts and protects the data inside its secure private cloud environment. The platform started processing the data automatically; it detected the data types and formats, ways to join different tables, and did all the data preparation and data cleanup steps that Dave would normally do manually. It then started to search for relevant insights, and also figured out how to best present these as intuitive graphs. After just a few hours of fully automated processing, the results were ready for Dave to explore.

Automated insights

Point Sigma's easy graphical insights and navigation allowed Dave to quickly get familiar with the different datasets and how Point Sigma had connected these together. Dave quickly noticed a number of patterns that he was already familiar with, which assured him of the data quality and the platform's ability to pick up relevant insights. Even though he knew these correlations existed, seeing them clearly laid out in a graph proved very useful. For example, the loan amount the customer requests increases with metrics related to the company size, such as debtors or liabilities, but for the latter it plateaus after a certain size. Dave explains: “When including these factors in a machine learning model, it is not only important to know that there is such a correlation, but also what shape it takes. Point Sigma's intuitive graphical presentation made it easy to see these correlations quickly.”

Debtors by Amount Offer
Amount Offer by Total Liabilities


Testing hypothesis in seconds

The absence of an expected correlation can be as insightful as the presence of an unexpected one. Using Point Sigma's intuitive keyword search, Dave tested a hypothesis that Funding Circle's customers tend to be located in more affluent areas of the country, but found this is far from the case. Funding Circle's sophisticated credit models are able to select the creditworthy businesses regardless of the area the business is located and their lending is nationwide. Funding Circle's 2021 Economic Impact Report outlined their impact across the UK in lending to SMEs in each parliamentary constituency. This consultancy-produced report explained that Funding Circle's lending is over-indexed in regions identified as having the largest investment gaps in the Government's Levelling Up White Paper: the East Midlands, Yorkshire and the Humber, and the West Midlands. On this dimension, Point Sigma was able to show clear evidence of the same insight after only a few minutes of setup.


Index Multiple Deprivation by Zip / Postal Code



After spending just a few minutes exploring and using automatically generated insights, Dave was able to form a detailed understanding of many aspects of Funding Circle and its customers. Dave adds: “This analysis would not have been possible without Point Sigma. There are so many things to take into account, like formatting, transformations, cleanup, data quality issues and many more and the project would quickly become intractable. With Point Sigma, we quickly got a systematic understanding of all the relevant aspects of the Doorda data to our business.”

Summary of outcomes

  • Great timesaver: just upload your data, and let the platform find relevant insights
  • Identified insights to improve understanding and decision making
  • Systematic exploration of insights across vast datasets
  • Easy visual exploration of statistical hypotheses
  • Confirmation that the data reflects the key aspects of expected business operation