Using Acquisition PV: Digital Affiliate Partner Evaluation
In the world of digital marketing for consumer lending, third party affiliate sites offer several advantages to both consumers and financial services providers
Affiliate sites provide numerous credit offers at one place to consumers, while complementing providers’ own marketing channels to engage with consumers
However, for financial services providers, identifying the right affiliate partner who provides quality consumer leads at a profitable bid is challenging
Building an internal valuation estimate for accounts booked through these sites is a first step for determining which leads will be profitable to begin with
Analyzing performance data of these accounts over time, continual valuation model refinement and sound application of data science can pave the path to best-in-class valuations driven acquisition strategies for lenders
The emergence of digital marketing channels continues to revolutionize every industry, and consumer lending is not exempt. Complementing traditional acquisitions channels with a strong digital presence is paramount to maximizing potential customer exposure. Third party affiliate sites in particular offer several advantages to both consumers and issuers. Consumers see numerous available offers in one place and can select the product with terms best suited to their needs, while issuers gain access to a broader pool of applicants than they could through their own marketing channels, paying only for those who apply and get approved.
Onboarding a new affiliate is an intuitive way to gain exposure to new customer segments and ultimately grow bookings. Comparing dozens of affiliates and determining which best suits your business needs is more challenging, and so is understanding the quality of leads provided by them. The definition of credit segments varies by affiliates, in fact the “Good Credit” category varies by up to 80 FICO points across sites. Moreover, the payout per account is set by the issuer ranging from less than $50 per account to over $500 in some segments, creating a complex strategical decision framework with multiple moving parts.
Definitions of Credit Bands across Affiliates
To further illustrate the problem, let’s assume through competitive analysis and testing an issuer has obtained the following data on two affiliates to consider partnering with:
$75 payout/account, Customer leads with FICO normally distributed from 600 to 680
$100 payout/account, Customer leads with FICO normally distributed from 600 to 700
This leaves the business with a clear choice: does it want to pay more for better application quality? Unfortunately, the optimal decision is not so clear. Is increasing the top end of the FICO range by 20 points worth paying an additional $25 for each account? The solution here lies in understanding the value of each account – the total interest revenue, losses, fees, rewards cost, interchange, and more over the life of the account. To discover the relative value of the accounts, we could book a handful of accounts from each affiliate, track their performance over time, and ultimately calculate the net income each generated. The issue with this approach of course is it would take many years to accurately measure the value, at which point the current marketplace will be notably changed and the original opportunity will have evaporated. To make an optimal decision in this situation we need to be able to accurately predict the lifetime value of new accounts using the data available when customers apply, which we can do with an acquisition present value (PV) model.
This statement could immediately follow the questions you are posing. Also at this stage, we should expand on what constitutes the value of each account: losses, finance charges, interchange and rewards costs at a high level, and explain how all of these come together to create an estimated value. This then opens the door for testing.
An acquisition PV model leverages historic performance data and assumptions about future performance to predict the value of new bookings. In the example outlined earlier, we would use the acquisition PV model to determine whether the improved selection increases PV by at least $25, thus covering the increased cost to acquire. However, this is just one example of leveraging data science to develop a robust acquisitions strategy. Other applications include evaluating the efficiency of increasing payouts to improve selection on current partners, determining whether to increase application volume by offering better product terms, or comparing different affiliates and focusing payouts on better performing affiliates.
We previously discuss testing as a challenge but are suggesting using a model which presumes knowledge of performance assumptions (which would have come from either historic testing or performance analysis), so will need to reconcile both ideas. One answer is to leverage already existing data within the bank and/or make informed projections of performance drivers.
Building an acquisition PV model can seem like a daunting task, and rightfully so. Model development often resembles an art as much as it does a science, requiring a robust understanding of both data science and the complex dynamics of the issuer’s lending business. The process begins by analyzing historic portfolio performance and grouping accounts into segments with distinct performance, and then using historic performance in those segments along with an assumption regarding future performance to predict the value of new accounts. While this first step advances a lending business dramatically, strengthening the model through continued monitoring and testing pushes a business to best in class. Continuously improving the underlying assumptions in the initial model by testing second order relationships and regularly monitoring actual driver performance against predictions increases the accuracy of the model, allowing the business to make progressively more precise strategic decisions.
Developing an accurate acquisitions PV model can be intimidating, but is table stakes in the fiercely competitive digital marketing space. Knowing what each customer is worth to the business allows an issuer to optimize targeting and payouts, acquiring the customers they want while minimizing cost.
AQN’s team of data scientists and consultants possess the rare blend of talent and practitioner experience necessary to build and implement a valuations framework for any lending business of scale. Our team works with clients to transform their data into an actionable format, understand the business’s unique needs to tailor customer segments, and build predictive curves based on past performance along with assumptions about future performance. With the underlying statistical analysis in place, AQN then develops a framework which allows clients to clearly understand unit economics and view KPIs for each segment and game “what if” scenarios for new strategies. We can also help set up, or manage, an ongoing testing strategy and monitoring process to ensure our clients get the most value out of the model for years to come.