Let's Look at Data At Leading Financial Tech Startup
LendStreet is a consumer financial health platform that offers loans to assist Americans who are deeply indebted. The product is quite complex and what we build internally is the entire customer experience as well as administrative software, which works with treasury partners to fulfill loans. LendStreet has a third party that services their loans. Because of its complex financial operation, LendStreet accrues a massive amount of data across many different data sources. The company gathers data from the team's owned operations as well as from third party partners.
Taj Sangha, LendStreet’s chief data steward and Product Manager has been at LendStreet for four years, joining the company right after college, where he studied mathematics.
He’s had several roles at LendStreet including customer support, compliance, operations and more. In the last two years he’s focused on LendStreet’s data and analytics program.
Taj joined us today to talk about how LendStreet came about to choosing Panoply as its solution to centralize and manage all their data in one place.
What decisions are made with data at LendStreet?
Taj explained that LendStreet’s entire business is informed with data from top to bottom - they use data in marketing to understand campaigns better, for example, using Google Analytics or Wordpress Analytics to better understand how visitors are interacting with their site and content. They also have product analytics tools to see how borrowers are interacting with specific pages or resources, and they track the entire marketing conversion funnel of customers from submitting loan applications to signing documents to actually getting funded. LendStreet has a need to see how all that activity is progressing to maximize profitability. Taj explains that they constantly track levels of engagement at each stage of the funnel and look for where any fall-off occurs. In addition, they use data about product performance to inform their product decisions.
Lately, LendStreet has been going through a massive project that changes the way they underwrite loans and are looking into creating views against data, which is not housed in their production database. In addition, they’ve also loaded massive amounts of historic data - this has been very helpful to better build their new underwriting model.
One thing LendStreet is always trying to track and predict is a specific loan’s likelihood of default compared to other loans of this type. To this point, they built custom attributes that they never were able to before. In using various types of banking data, they obtained access to a treasure trove of data on which to build an underwriting model.
“I can’t think of a place in our business that we’re not making data-driven decisions - based on data collected and optimized in Panoply,” Taj explained.
Excerpts from our interview with Taj
Describe your experience before and after having your data stack in place?
Prior to setting up Panoply, if we needed to collect and analyze LendStreet-generated data - we queried our production PostgreSQL database and some of the JSON columns would completely bring our database to its knees. For external/third-party data, we receive files on a daily/weekly basis - and that data was stored in Excel spreadsheets. Our ongoing data management was cumbersome and proper analysis and detecting data anomalies was close to impossible.
We also had this laborious process of generating our business intelligence graphs.
Now, with our data stack, we don’t have this problem because all of our data is query-able and ready for analysis. We use Tableau on top of Panoply - and this data stack is ingesting data and feeding itself - I don’t even have to worry about data collection and analysis any more. It’s a night and day difference.
For our team, having all data optimized in one place means my team and I can now go into Panoply and create views and make sure the data itself is constructed the way a business or operations user can consume it and make vital decisions based off of it.
For example, we have one field duplicated across two tables and a business user won’t know which field to pull data from - with our views in Panoply, we can ensure users are reporting on the right data.
We can now consolidate our work and have a separate environment just for analytics.
How did you decide upon Panoply as your data warehouse of choice?
Honestly it was a revelation to learn that Panoply existed! We evaluated other tools such as BigQuery, Snowflake and Redshift - and without fail, with each one of these, we’d have to dedicate engineering resources to setting up and maintaining the warehouse and ensuring our ETL wouldn’t break. Somehow we discovered Panoply and we were so relieved to see a solution existed that would manage all our data with little to no ongoing maintenance and fuss.
Also, we enjoyed Panoply’s upfront and predictive pricing and amazing support. We love the Panoply team and product.
What were your business intelligence requirements when evaluating your data stack?
We needed a central store. We need to have ETL that’s being managed automatically- not by custom scripts we had to run. Lastly, we needed our analytics and data management solution to not need any support from LendStreet’s engineering team. To be honest, no other provider checked all these boxes as well or as fluently as Panoply.
We thought initially we’d set up another PostgreSQL database and use an ETL provider to pull data into it, but once we discovered Panoply, that desire fizzed out really fast.
Which sources did LendStreet collect and combine in Panoply?
We bring many data points into Panoply, including:
- Google Analytics
- Accounting data
- Third-party data that originates from nightly files into S3 and therefore into Panoply
Which teams use Tableau/Panoply and how do they use data?
LendStreet's Finance Director and Taj are the most active data consumers at the company. They have regular meetings each week where they pull up data relevant to their meeting.
These are some of the questions they ask of the data they normally collect:
- How many applications are we getting for this time period?
- Are we on track for our monthly origination goals?
- How are our different partners performing in terms of credit quality?
- How are we performing in terms of delinquencies?
- Our business development director gauges whether our partners are producing leads and loans according to their goals.
- Sales/underwriting teams uses data to track turn times and times to process applications
LendStreet tracks all of the above in real-time.
What have been LendStreet’s big wins in terms of data and what metrics can you tie to those wins?
As a company, LendStreet has been able to improve their credit rating, because they are able to approve more applicants and have a much lower loss rate because of their predictive data model.
What’s Taj’s favorite Panoply feature?
“This has to be Panoply’s customer service combined with the ease of setting up a new data source. ETL ‘just works’!!! “Any idiot can do it", says Sangha
Thanks for joining us and sharing your company's data journey, Taj!