The manifest destiny approach is easy. For instance, low funds value, reduced overdraft fees and declining debit interchange are making your checking accounts only marginally profitable. As a result, you are marketing them less intensively.
So you have your pricing expert determine the fee levels that will cover the recent shortfalls in revenue. For instance, $10 per month for new accounts. (If you're lucky, your expert will be incisive enough to take some account of the fee's effect on acquisition, retention and use — which is guesswork.) Then you add a few exemptions, and you roll it out. This is the way it had to be.
It's simple, simple-minded and arbitrary.
After all, many alternatives exist for structuring fees. Should there be a minimum average or a minimum low balance for exemption? And/or possibly a minimum number or dollar value of debit transactions? Should there be interest or rewards? There are at least 12 axes along which checking account fees can be denominated, and there are many kinds of account benefits, limitations and rewards.
But the old-fashioned test-and-learn approaches that require building software to deploy in test branches can take a year or more and consume a lot of money and management effort. A fast test-and-learn approach gets more results using far less time and investment.
A broad stream of people who are soon going to open checking accounts can be found on the Internet. Most want a bank with a convenient branch, not an Internet-only account.
If you have alternative account versions to test — A, B, C and D — and you can deal with them one by one, then show a prospective depositor an account with the A set of terms compared to some competitor's offers, a second prospect an account with the B set of terms compared to the same competitor's offers, … and when you get to the fifth prospect, the A set of terms would again be offered. In each case the prospect clicks to choose.
This will show what share of the people offered A prefer it to the competition, the share that preferred B and so forth. Putting these data together with the likely balances and use for each prospect lets the bank project the top- and bottom-line results for each of the four alternatives.
In a few days, you finish that experiment and move on. The next step might be to try another set of four alternatives, one of which would be the "champion" from the first round. Suppose in the first test, a product with a monthly fee and low debit rewards soundly defeated even product versions with much lower fees but no rewards. Then, in the second round, you might try a version with a higher fee and higher debit rewards — even rewards equal to and integrated with your credit card rewards.
This rapid, dynamic testing, hill-climbing to the best account structure, is possible only if you do not have to build software to deliver every test version of the product and do elaborate in-branch training for it.
A vendor can perform these tests on an unbranded basis — talking to people now looking for checking accounts who have characteristics similar to your customers. Prospects will supply demographic and use data, and they will chat or make an inbound phone call to facilitate in-depth discussion. The branding effect can be measured separately if needed.
