The economics of small business lending did not make sense for a while. Startups and fintechs in the 2010s and the beginning of 20s had to learn this the hard way, by givingby trying and giving out loans that were never profitable. The old adage of lending it is easy to give out money, it's harder to get it back couldn’t be more true.
Despite this, a couple companies broke through. Wayflyer, Founderpath and Capchase in the 20s are interesting businesses to look at. They enable access to capital for those who could not previously do this.
So what do they have in common?
Efficient back office and data processing. The “magic of underwriting” and “secret sauce” so many new lenders promise really lies in the operational efficiencies of qualifying smaller customers in bulk, without having to hire hundreds of people.
If you cannot do this and understand SMB-s at scale, they do not make sense for you as a borrower and they never will.
There is something else the above mentioned companies have in common. They are all Ntropy customers.
We built Ntropy’s data enrichment API-s to allow underwriters evaluate businesses quickly from the source of truth, bank data. Even if your accounting is not up to date, your bank feeds always are.
Today we are taking another step forward in this direction.
We are launching Insights, an easy-to-use analytics tool for business lenders to instantly recreate financial statements, P&L and cashflow, from the source of truth, bank transactions enabling them to scale their lending volumes without hiring more analysts or adding more processes.
Why did we build Insights?
Business lenders are some of our top customers as our enrichment solutions let them better understand the risks associated with companies they were underwriting.
However, after spending considerable time listening to lenders such as Uncapped, Liberis, Founderpath, Wayflyer and Aqeel, we discovered that their teams were still spending over 80% of their time reconciling, mapping and structuring financial data sources with different structures and turning them into a standard P&L view they can read and evaluate.
The data they receive from prospective borrowers includes bank data, financial statements, accounting information and payments and commerce feeds and is sent in a number of different formats (PDF, API and spreadsheets) with no consistency across customers or geography (data model, standardise, reconcile).
Structuring and making sense of this data is foundational for them to be able to effectively underwrite. Whilst we are solving a big pain point by cleaning, standardizing and categorizing bank transactions, underwriters still need to do a lot of manual work before analyzing it.
Given the language models we have been working on to understand financial data from a variety of sources and in a wide range of formats , we believe we are uniquely positioned to help and solve this problem further.
Whilst a lot of the buzz on the use of LLMs in finance has been on the front end, we believe there is a much bigger opportunity to revolutionize the often overlooked back office. The reason financial services are expensive and not accessible to everyone is because running an FI is operationally heavy and costly. KYC, KYB, Compliance, Risk : all of these involve human heavy processes. These can often be a great fit for language models, only cheaper and orders of magnitude faster.
Why is SMB lending important?
SMBs are the lifeblood of the global economy, frequently accounting for 60%-70% of jobs in most countries and more than 50% of employment worldwide. Those numbers are even higher in developing economies.
However SMBs are chronically underserved when it comes to financial services. Estimates of the global funding gap for SMBs between $2trn and $3.2trn with a £22bn gap in the UK alone. Data from the Federal Reserve found 44% of US SMEs who applied for traditional bank loans last year received none of the credit they asked for.
However underwriting SMBs presents a problem for lenders. The data on small businesses is often lacking in quality and consistency. It is often fragmented across different software systems like accounting, banking, payments or commerce platforms which all told, makes underwriting a very manual process. Fortunately, many new fintech lenders have entered the market to try and help plug this capital gap using new sources of data.
However the manual process to utilise these different data sources, a process known as “spreading”, is hampering the process at plugging this credit gap.
How does Insights work?
Insights takes raw bank transactions, statements, csv-s and outputs clear company summaries that are easy to examine and model on regardless of the size or the industry.
This is the natural evolution of Ntropy’s enrichment product as the majority of our customers were using the enriched view to understand companies better: revenue, COGs, operating expenses and other P&L based categories being the most demanded fields.
Many in the market have business lending dashboards. Yet, without highly accurate enrichment, additional insights are not possible or do not make much sense. Ntropy’s AI provides Insights with the most accurate enrichment to build on top of.
Existing solutions that lenders we have spoken to use still require lots of manual corrections they barely solve the problem.
Insights on the other hand is built differently. Ntropy customers get the benefit of the best in class language models fit to understand financial data. Imagine the power of GPT4 or Palm 2, or the next generation of models, specifically fit to analyze your data at a fraction of cost and latency?
Who is Insights for?
Insights is built for business lenders with complex underwriting processes that utilise multiple data sources.
Based on our interactions with customers, as mentioned earlier, underwriters spend 80% of their time structuring different data sources and data types before reconciling it all against the source of truth, bank data.
For underwriters, this process of data transformation is incredibly tedious and time consuming and becomes the bottleneck for making quick underwriting decisions that borrowers need.
Whilst for Heads of Risk, the manual nature of the data transformation brings inconsistency to the process and limits their ability to have total control. Providing a single, centralised process for recreating financial statements ensures Head of Risk have higher confidence in their underwriting decisions.
In order for lenders to continue to grow, they need to increase lending volumes and there are two ways to do that
- Hire more underwriters
- Spend less time making each decision
The first option is expensive and not possible in the current business environment. The second option, without the right tooling, can lead to higher risk if underwriters spend less time reviewing a borrower’s financial data in order to move on to the next borrower.
With Insights, lenders can scale their underwriting without increasing headcount or increasing risk through removing the bottleneck of manual data reconciliation, transformation and summarisation.