In our previous post, we saw how overlooking very specific metrics can result in missing important business incidents. These data “blind spots” – the inability to effectively monitor data – is a major flaw of traditional BI analytics tools, and this flaw prevents them from producing actionable insights in real-time, which is an absolute necessity.

The example we used was that of an API change which broke payment processing of orders made with American Express credit cards for an ecommerce company, a scenario that illustrates the tight link that now exists between ecommerce and fintech. Just like in ecommerce, there are many specific metrics in this industry which can be overlooked if companies monitor data inadequately, and when a business incident occurs, those overlooked metrics can contain the clues needed to put the big picture together.

In order to effectively monitor data, ALL metrics must be taken into account

Fintech is the transformation of financial services by technologies such as big data, mobile apps, and machine learning. Basically, it’s equal parts finance and technology, and that means that any company in this space needs a solution that can continuously monitor not only business metrics such as daily revenue, cash on hand, and number of accounts, but technical metrics as well (e.g. page load times, rates of timeouts, number of simultaneous connections, percentage of database memory used). Any code breakage can stop someone from getting approved for a credit card or prevent someone from getting paid.

To lend or not to lend, that is the question

What types of “blind spots” can occur when a company’s data monitoring tools fail to provide real-time insights into all individual metrics?

Online lending has gained popularity with its higher efficiency in offering credit to consumers and small businesses. Nevertheless, a fintech company that provides an online marketplace for such loans, matching lenders and borrowers, bears higher risks compared to traditional bank consumer loans due to insufficient credit checking, inadequate intermediation, lack of transparency and the inherent financial status of typical online borrowers. Loan application decisions are made automatically through electronic data-driven algorithms. Therefore, credit risk prediction and monitoring is critical for the success of the business model.

By monitoring the outputs of the credit risk prediction algorithm at a deep level of granularity (e.g., by city or type of users, etc), an anomaly detection solution can identify when something is wrong with the risk prediction algorithm — a situation that can happen because of erroneous inputs (due data integrity), a bug in the algorithm, missing data, etc.

This is just an example of the type of specific signals that could get overlooked by traditional BI tools that monitor data. Old-style BI tools only cope with part of a problem, only focusing on the data they think they might need.

Monitor your data in real time or face the consequences

When there are hundreds of individual metrics for each of the thousands of users on such a platform every day, only a scalable solution which is able to quickly connect all the dots will suffice. Unfortunately, traditional tools that monitor data like KPI dashboards can’t keep up because they often work with historical data, not real time information. And when the many moving parts in on online lending platform can change frequently,  stale data could undermine the business model and impact profits. Too many fintech businesses may end up waiting too long to catch serious, yet subtle, business issues.

It’s easy to know right away when something crashes hard. When a problem is limited or the impact is more subtle to a top-level KPI, it can take days to find out and remedy it. Meanwhile, customer satisfaction can drop,  and so can revenue.

If a partner offer is suddenly under-performing, you need to know now. If your customers are not leaking, but pouring out of your funnel due to increased page latency, the sooner you can identify the problem, the sooner you can get your own company’s financials back on track.

Topics: AnalyticsArtificial Intelligence
Subscribe to our newsletter and get the latest news on AI analytics

Written by Ira Cohen

Ira Cohen is chief data scientist and co-founder of Anodot, where he develops real-time multivariate anomaly detection algorithms designed to oversee millions of time series signals. He holds a PhD in machine learning from the University of Illinois at Urbana-Champaign and has more than 12 years of industry experience.

Get a Taste of Anodot
footer background