New Strategies for Predicting Default

We recently hosted a session at the Lend360 virtual conference. Here’s a recap.

It’s been a conference season unlike any other. We sincerely miss seeing all of you face-to-face but want to thank everyone who attended our recent virtual Lend360 session.

We had a great time sharing some thoughts about our new Plat.AI indexes. Your feedback was very useful, and we look forward to helping many of you get started with these indexes.

If you missed the conference, then read on for a synopsis of what we covered during our session.

Responding to a Crisis

Man fixing a downward trend.

The main focus of our session was looking at custom data solutions and proprietary risk indexes from Plat.AI. Specifically, we talked about our indexes built to mitigate default risk caused by COVID-19.

Many of you are concerned about higher default rates and consumer risk in general as a result of the closures and shutdowns. It is our desire to help you navigate this.

Plat.AI is our suite of data services. There are custom models as well as proprietary data indexes. These could be implemented very easily through our Plat system. They are turnkey solutions. You can integrate these and start applying them to your lead buying and your marketing campaigns right away.

We could literally have these up and running within a day or two and help you mitigate some of that risk in this environment today.

Our Unique Approach

One rocket blasting off while others do not.

As many of you probably did in March and April, a lot of our lender clients completely paused their campaigns because of the uncertainty. We didn’t know what would happen or how long it would last.

Here we are in November with the holidays coming up, and the country is sort of in the same place. However, our industry is evolving.

Our clients are asking about the stability of consumers: “Hey, is this consumer going to have a job two weeks or a month from now? They work for cruise lines or restaurant businesses.”

We have answers. Our indexes are really a response to the concerns of our clients.

How Plat.AI Can Help

Face made out of code.

Plat.AI delivers real-time predictive analytics to the financial sector. We design, build, and deploy various machine-learning models to tackle problems and issues. Specifically, we build scorecards and risk-assessment models. We have done this for more than four years now, working with over 400 different lenders. We have more than 500 deployed models. Most of those are custom models with specific targets.

Why Plat.AI?

In this age of machine-made lending decisions, lenders are constantly looking for additional tools to better manage their risk. The common problems they are facing are accuracy, latency, security, and so on.

Plat.AI provides fast and accurate real-time predictive analytics. More importantly, the service includes maintenance, so the models are not just a one-time deal where you build a model, deploy, and forget about it. Data scientists constantly update, upgrade, and tune-up those models to make sure they always stay on target and give the best performance possible.

Our Inspiration for Creating Stability-Assessment Indexes

Think about an Amazon warehouse worker versus a bartender who works for a cruise line. How stable are those employments? Or, what about jobs within the same industry but in different states? Let’s say a movie theater. Different states are responding differently to the current situation. Therefore, employment stability changes as you move from state to state.

This was the motivation that started our employment stability indexes. We are looking at major factors that can affect stability, like the consumer’s industry. These are machine-learning models that are producing risk scores for our clients’ marketing campaigns – or any other campaign that requires a real-time decision-making process.

Four Game-Changing Indexes

Using the risk indexes in model-building highly increases the predictive power of our models. These are four of Plat.AI’s most-used indexes:

  • EDSI – Employer Default Stability Index
  • ESI – Employer Stability Index
  • ESSI – Employer State Stability Index
  • CSI – Consumer Stability Index

EDSI, ESI, and ESSI are plug-in-and-play services. These are already-trained, ready-to-go models. Within a day, you can integrate them with your existing platform.

An In-Depth Look at CSI

A consumer staying stable.

Some lenders need customized solutions to go after various targets, like bad accounts, good accounts, sales, you name it.

Speaking about the use cases of the Consumer Stability Index, one simple example could be targeting sales. Let’s say you’re receiving 100,000 leads, and you spend a dollar in order to return a response whether to accept or reject a lead. If you are receiving 100,000 leads and you end up buying 5,000 of those, then you are paying $100,000 for 5,000 leads.

We can train a model targeting sales and cut 30% of the volume. In this case, the client would spend $70,000 instead of $100,000 while not losing any sales. The client would spend 30% less and end up with the same number of sales.

Setup is fast and does require passing us any PII. These solutions are developed within weeks – not months or half a year like many are used to. They can be delivered through a very simple integration process that we have. We have integration with underwriting platforms, so we provide those scores.

Why Data Science is Part of Dot818

We realized about four or five years ago that this isn’t a game of only providing leads to our clients anymore. The days of being just a lead provider and slinging leads over the fence to your lenders and other clients are over.

As a reaction to the market, we implemented a data team to implement custom models that target both conversion and long-term loan performance.

We now provide a unique service to the lending community – end-to-end agency services. Our data and technology services allow you to be much smarter than you were even two or three years ago.

We help our clients buy leads more effectively. Please contact us if you have any questions about getting more out of your data.