In another article on AI for Lending we referenced the fact that AI excels at helping lenders solve the fundamental information asymmetries in their business. The online lending ecosystem is very complex and has many players with different roles. In fact for lenders, acquiring customers is a huge pain point. As a result, there are many companies that specialize solely in acquiring and routing leads to potential lenders, and they might be called lead brokers. AI can help with one of the fundamental problems lead brokers have: lead optimization.
Lead brokers have their own set of problems which can also be thought of as related to information asymmetries. Typically a lead broker sits in between lead sources (sometimes called publishers or affiliates) and lenders, or often even other lead brokers. A useful analogy may be to think of a financial market maker that hopes to buy or acquire leads (they always have some cost) as cost-effectively as possible and route the lead to the best lender eventually. Here “best” is defined as the lender who values the lead the most. In this way, the lead broker is trying to make the market for leads efficient, by sending the customer to the lender who values the customer the most in a competitive marketplace. The competitive marketplace of lead brokers helps insure that customers themselves get the best lending terms.

So, for a lead broker, the task is to estimate, with imperfect information, which lenders (or other lead brokers) most value a customer. The broker can know many things about a lead such as its source, a variety of characteristics about the customer (some of which may be inaccurate), etc. The broker may also have a lot of (also imperfect and noisy) information about the lenders and other brokers. The lead optimization task is then core to the business of the broker. Lead optimization is the process of deciding actually what to do with a lead most broadly. Which leads should be offered to which lender or broker? At what price? In what order? The decision problem is very complex and all occurs in an economy and ecosystem that can be rapidly changing. It is very easy to mistake noise for signal in making these decisions.
Determining signal from noise is precisely the central task of AI in lead optimization. Many lead brokers have grown businesses that began with fewer leads, a less complex ecosystem, and far less scale and data on customers. Industry knowledge and intuition have worked well to define the process of lead optimization.
But as fintech and lending, in particular, grows online, there is a need for a systematic algorithmic AI / Machine Learning-based approach to lead optimization. AI is perfectly suited to help remove outdated intuitions and hunches and let the data speak objectively. AI for lead optimization can dramatically increase revenue while at the same time driving down operational costs by streamlining legacy workflows and processes that manage the core lead optimization process with unknown efficiency.
A systematic AI / ML approach ensures that all decisions are also experiments with well-defined answers. These experiments lead to ever increasing efficiency, and point the way to new opportunities for growth by highlighting exactly where the weaknesses and underutilizations are.
A systematic AI approach eliminates the complexity and fog that can surround more manual and inertia-driven processes based on human decision-making in which it is unclear what would have happened if different decisions had been made.
The systematic AI / ML approach makes use of cutting techniques such as multifactor A/B testing, multi-armed bandits, reinforcement learning, etc. that are core techniques used by the largest internet companies in the world such as Facebook, Google and Amazon. These large companies, which have massive advertising businesses, are in many ways lead brokers themselves. They too have to decide how to most efficiently route a lead / customer to a company that wishes to serve that customer, among many such companies. While those companies literally spend billions of dollars on the systems and AI talent needed to create and sustain this core activity, they do not offer their technology, which is obviously extremely effective given their profitability, to smaller companies.
Thus, there is a great need for AI lead optimization algorithms and technology based on the latest cutting-edge AI research within smaller companies who would benefit and gain great competitive advantage, but who would find the creation and sustenance of an in-house AI capability cost-prohibitive and extremely difficult and uncertain.