The Consumerist is great fun, but sometimes they, or their commenting public, kind of miss the mark — as they do in this story about a declined check at a K-Mart.
S. wrote a check at Kmart earlier this month and it was denied. No reason was given—just “denied.” It turns out a separate company, Certegy, made the decision, so S.—who writes, “I’ve never had a bounced check”—tried to track down someone at Certegy who could tell her what was wrong with her checks.
Cry me a river. First, you’re writing checks in 2008? Seriously, WTF?
More seriously, though, what the whiner ran into was a risk-based turndown. This post describes something that’s pretty common in the check verification industry. I used to work in software development for Telecheck, but it’s been a long time. The basics of the business are probably unchanged, though.
Back then (late 1990s), TCK had two main products for merchants:
- Their flagship guarantee product; and
- A verification-and-collections product that was cheaper.
Guarantee cost a percentage of each processed transaction, but it meant that if TCK issued an approval code, then the merchant was covered — if the check went bad, Telecheck paid him anyway, and collected the money plus the bounce fee on the back end.
Verification-and-collection was just what it sounds like: they’d run checks at POS, and approve or deny against the same database (with some differences), but bad checks were just bad checks. The merchant would only get paid for them if TCK managed to collect, minus a commission.
So, when a check is run through the system, the first thing that happened was a search for any actual negative data. For TCK, you’re negative pretty much only if you actually owe TCK money and/or the bounce fee. People who bounce checks and then pay TCK later after a paper notice are TCK’s favorite people (think about it), so neg-data turndowns only happened if you had an open item; that turndown was called a Code 4. Owe Telecheck money, or owe a Telecheck verification client money? No checkwriting for you. Don’t? No neg data. Knock yourself out.
That was the end of the story, as I recall it, for verification customers.
However, with guarantee, actual risk analysis turndowns came into play for people without neg data. As a boss of mine used to say, there are two kinds of bad checks: people borrowing money, and people committing fraud. The former are collectable, and the latter aren’t. The trick is knowing which is which.
To try and eliminate fraud, they used scorecards. There used to be a single checkwriting scorecard, across all SIC (essentially, type of merchant) codes, published by (I think) Fair-Isaacs. TCK developed a whole bunch more, since it turns out that LOTS of factors correlate to the relative riskyness of an unknown checkwriter, for example:
- Men are riskier than women
- Younger people are riskier than older people
- New accounts are riskier than established accounts
- Low check numbers are riskier than higher check numbers
- Checks written at the end of the day, or in particular Friday afternoon, are riskier than checks from earlier in the business day/week.
- Etc.
Obviously, too, some merchandise is riskier than others. Subwoofers are risky. Carrots aren’t. All this intelligence — and there was a half a floor in Houston full of very smart people doing the analysis behind this — came into play only for guarantee customers, since it was actually Telecheck’s money getting risked there.
Getting a risk-based turndown from TCK meant you looked too dicey for them to say, absolutely, we’re gonna cover this check for the merchant. The merchant could, of course, decide to take it anyway (but would get no guarantee), and will certainly suggest another form of payment, but TCK just doesn’t want any part of it. For Telecheck, the risk turndowns were Code 3.
Now, back then, some other companies were trying to also do risk management turndowns, but they’d do stupid things like simple velocity turndowns (“no more than N checks in Y period of time”), which is mathematically indefensible, or even simple cumulative price limits (also stupid). TCK had LOTS of years of actual POS data to draw from to create valid predictive models, which is what made them the higher-end provider back then.
So, at the end of the day, risk turndowns are just that: risk management. I don’t know anything about these new companies in the check verification market, and (as I said) my TCK insider knowledge is a decade old, but back then the whole code 3 thing wasn’t surprising or weird to me. It seemed like good business based on the inherent riskiness of checks and the inventive product (for the time) that TCK was selling (guarantee). Sure, people whined about it, but TCK wasn’t and isn’t in business to make checkwriters happy. They’re in business to make sure POS checks are as safe as possible for their clients.
Endnote: Since college, I’ve never written POS checks, even when I worked for TCK. Too much trouble. Amex uber alles.