Too often solving one problem creates others. Take Check 21. By allowing check conversion to electronic form the sensible idea was to reduce costs and speed processing. The problem is that with physical checks, image replacement documents, ACH transactions and remote deposit capture all now occurring, there are many more check pathways into the bank, and so many more chances that some of those checks are duplicates. It is, in the words of Michael Manila, director of The Payment Group, a consulting practice, “the perfect storm of opportunity for duplicate postings.”
Frank Stokes, president of technology systems at Conix Systems, says, “Most banks don’t realize how big a problem they have.” The industry average for duplicate checks is 120 per million, and while that might not sound like much, Stokes says a small-to-medium sized bank could easily process 2.5 million items per day. Since typically each duplicate check takes $70 to remediate, those costs add up fast. Calculate 21 working days a month times 12 months and the annual cost comes to nearly $5.3 million. Duplicate checks are also irritating and confusing for customers.
One bank that recognized this problem early on was BB&T Bank, which decided in January 2007 it needed to develop technology that could detect these duplicate checks before being processed. Tim Dillow, svp regional operations support manager, says the bank’s 3.5 million processed items per day gave it ample financial incentive to head off duplicate checks. Dillow says the bank considered taking on the technology project itself, but quickly realized it needed a technology partner and struck an agreement with Conix. “Conix really listened to what we wanted, and once implemented, when we realized we wanted more, they listened to us again to create enhancements and make the solution user friendly. We really feel we have a partner.”
A big problem with the first-generation solution from Conix was too many false positives. Rebate checks, mortgage drafts, really any recurring checks with identical amounts were flagged by the system. Since then, however, better filters and business rules have been incorporated into the solution, called Dupe Detective. For example, in July BB&T processed 3.5 million checks per day, and Dupe Detective initially flagged about 15,000 as suspicious during the month. From there the business rules took over and confirmed that 15 percent of those 15,000 were in fact duplicates. BB&T planned to go live in August.
Conix, which owns Dupe Detective, is starting to license the solution to banks for between $75,000 and $1 million depending on the size of the institution. Dupe Detective cannot stop duplicate checks from being sent to the institution, but it detects them on day 1, not day 2, by which time so much damage has been done, i.e., actually posting to a customer account. The cost of unwinding a duplicate check day 1 is not insignificant, $20 to $25 per check, but far less than the $50 to $75 it costs on day 2, Stokes says.
“We look across channels—IRDs, checks, ACH and e-checks—and we’ve integrated Dupe Detective with balancing systems and adjustments systems at the bank,” Stokes says. He estimates that 20 percent of duplicate checks are paper, while the remaining 80 percent are in electronic form.
Manila of The Payment Group says that duplicate checks are obviously not new, but in the past there were fewer of them and, critically, usually involved smaller retail checks. What has jump-started the need for duplicate detection is that banks use IRDs for much bigger checks in order to reduce float and reduce exposure; thus, today, very large “controlled disbursement payments” are at risk of being duplicated. Instead of $50 retail checks, the bank is in danger of duplicating a $10 million commercial check. The stakes are higher. “The pain is bigger,” says Manila.
Any technology spend in today’s difficult economic climate is a tough decision, but Stokes and Dillow are certain that the return on investment with Dupe Detective is quick. Dillow says that a “very conservative” estimate is ROI within a year, but perhaps much faster if a severe duplicate check event were to occur.
For instance, if a cash letter were duplicated, an institution could literally face many thousands of duplicate checks in a single day. A cash letter is, essentially, a box full of checks, usually 5,000 to 10,000, which are grouped together and processed for convenience. When one of these is duplicated the results can be devastating. Manila recalls being called in to help a small bank in Montana cope with a deluge of duplicate checks that had been passed along by none other than the Federal Reserve. “They had 3,000 dupes in one day, and that really hurt them. They’re not staffed to handle all the research that has to be done.”
Other banks have systems to help detect duplicate checks, but often those systems don’t look at individual items, only duplicate files, Stokes says. What also makes the Conix system different is that it prevents the bank from sending out duplicate checks.
This outbound protection is a big concern for BB&T, says Dillow, given the web of trusted relationships it must maintain on a daily basis. Dupe Detective guards against “reputational risk,” he says.