Have you ever wondered how banks or credit unions can mail credit cards offers to you at your current address? How does the process work behind the scene? What impacts a campaign? Why do you receive the same offers from the same issuers but in different envelops? If you have such questions, I am here to pull back the curtain a bit by talking about the Direct Mail (DM) process in general, what impacts the success of a DM campaign and some less known details.
Direct Mail Process
Let’s go backwards from the moment you tear up a mail from an issuer. Whenever an issuer, whether it’s a bank or a credit union, sends you an offer, it must contain information on the offer as well as credit terms and conditions, as mandated by regulations. Inside a mail piece, some issuers include a postage-free envelop that customers can use to send back a filled paper application. Since each piece of paper is an expense, some elect not to send a paper application to save costs, especially during the times of supply chain constraints and inflation. In that case, customers can go directly to the issuer’s website, either by inputing the address manually on a browser or by scanning a QR code. They can also call customer care and apply on the phone.
Each mail piece carries an access code and a reservation code. These codes will be used to identify which offer is attached to an application. Normally, an issuer assigns a unique identifier to an offer for easy and transparent tracking. More on this later.
How do issuers know who you are and where you live? The answer is by working with credit bureaus. Credit bureaus like Experian, Equifax or Trans Union collect a lot of data on Americans. They have your latest address, your social security number, how many trades (mortgage, loans, credit cards…) that you have, how many with a balance that you own, so on and so forth. Issuers can work with these bureaus to pull the names of prospects for DM campaigns. The caveat is that in addition to a fee per name, issuers need to commit that they will send an offer to the names that they pull. Said another way, issuers can’t just call bureaus out of a blue and say: hey, I want to pull sensitive information of these people, but I don’t send them any offer.
Because of this requirement and the fact that each credit card is an unsecured loan that carries risks of losses, issuers must have a plan as to whom they want to send what. To answer these questions, issuers rely on their Credit Risk department, Marketing team and historical data. Credit Risk determines the risk parameters in which new acquisitions must fall. For instance, some banks are more comfortable with people who have little credit history than other banks. Some want to acquire folks with FICO less than 660 than others.
After Credit Risk defines the broad risk parameters, Marketing will work on the specific criteria and offers for a campaign. Each year, Marketing will conjure a campaign calendar that details how many pieces will be sent, when a campaign starts, which offers will be sent, how many applications and accounts can be expected. These details are determined with the help of historical data. Hence, the longer an issuer has been around, the more data it has to make informed decisions regarding DM campaigns.
What Impacts A Direct Mail Campaign?
The biggest factor is whether an issuer sends the right offer to the right audience. People have different preferences. Some don’t like complex rewards structures while others love to maximize rewards points. Some want to transfer balance to a card with a much lower interest rate while others just want to get a cash bonus for their activity. Issuers need to figure out who likes what and sends an appropriate offer. There is a big caveat. Credit cards are highly regulated in the US. Issuers can’t be caught being discriminatory towards any portion of the population. They can use certain behavioral traits as targeting attributes. What they can’t do is to use demographic elements. For instance, income, age, marital status, occupation or place of residence, just to name a few, are strictly forbidden.
Indicators from the bureaus such as how many trades a person has, the age of the oldest/latest trade, the total balance or how many delinquencies a person has can be used in a campaign. Issuers and bureaus themselves also try to use machine learning to build predictive models based on these legal attributes to gain an edge. The better the models are, the more efficient DM campaigns become.
Of course, offers with numerous benefits will excite prospects. As credit card is a fragmented business with a lot of competition, issuers cannot afford to come to prospects with bare bone offers. However, they must also think about their bottom line as rich products tend to be money-losers. Would you apply for a credit card with zero interest on balance transfer and purchase for 20 months, 2% cash back on everything and $400 bonus offer after spending $1,000 in the first 3 months? You likely would, but I can almost guarantee that the issuer of that card would not make a cent of profit. Hence, it’s all about finding that sweet spot between profitability and acquisition efficiency.
Additionally, mail design, paper quality, paper color and copywriting can contribute to the success of a campaign. I am sure you can recall seeing the same offer from the same company, but in different mail designs with different types of paper. Issuers conduct a lot of tests to see which paper or design can generate an extra basis point or two. Furthermore, the use of QR code can also help. USPS currently has a deal in which they will lower the postage expense if issuers use their Informed Delivery service. This is a numbers game. It’s all about finding those extra basis points in response rates.
Behind-The-Scene Details
Let’s start with promo/campaign codes. These codes are usually invisible to credit applicants. They are what issuers use internally to identify offers. In fact, each offer can have two promo codes: the parent code and the child code. The child promo code represents applicants that are upgraded or downgraded, depending on the setup of each campaign. For instance, anybody who applies for a Visa credit card and is awarded a credit line of more than $5,000 will receive a Signature card, instead of a Classic card. Signature cards carry more benefits and give more interchange revenue to issuers. The child promo codes for Visa campaigns are usually assigned to these “upgrades”. For Mastercard, the parent promo codes are given to the higher tier and the child codes are given to the “downgrades”.
What happens between bureaus and issuers? After an issuer finalizes a campaign’s strategy using criteria from a chosen bureau, the issuer will send the bureau such criteria and get back at least three files. The first file will go to the printing house and have some necessary information such as name, address, promo code or a unique identifier tagged to a mail piece called Solicitation ID. The file will not have people’s social security numbers. Nor will it have all the attributes that the bureau has at its disposal because the printing house doesn’t need to have such information.
The second file will go to the issuer and it has fields such as Solicitation ID, promo code and all the targeting attributes that the issuer and the bureau already agreed upon beforehand. These attributes will enable the issuer to analyze campaigns and see what can be the most predictive of success. Almost every issuer usually tasks its Machine Learning team to use multiple bureau attributes to come up with a predictive model so that it can use to generate more applications in future campaigns. Like the first, this second file will not have Social Security Numbers as Marketing or Machine Learning team does not need that kind of data.
The third file will also go to the issuer and be integrated into its decisioning engine. This file will have Social Security Number as Credit Risk and Operations will use it to make underwriting decisions. Of course, these teams don’t need all the targeting attributes as they are less relevant to them than to Marketing.
How do issuers deploy custom models? The answer is that issuers don’t “deploy” the models themselves. Credit bureaus do. After finalizing a model, an issuer will send the “formula” to its chosen credit bureau and the bureau will calculate the score based on such “formula”. The score will be appended to the appropriate files mentioned above and sent back to the issuer every campaign. The issuer will use the real performance data to validate the model and adjust, if necessary.
Every issuer must make sure that all models are in compliance with all lending regulations. Annually, the Office of the Controller of the Currency (OCC) conducts an audit to see if financial institutions comply with the regulations. Hence, every model must get approval from an issuer’s Compliance before deployment.
That’s all I have for today’s entry. I hope you find it useful. Drop me a line if you do or if you have questions.
Minh Quang Duong 