avatar
+2 2 votes

Barclays makes spreadsheet error in Lehmans firesale

Courtesy of my friends at Gartner, I see that Barclays has managed to drop a veritable clanger in acquiring 179 assets it didn't want in the Lehmans firesale. All due to spreadsheet error.

Those who know me will also know I have been railing against spreadsheet use for at least the last 10 years. The most egregious errors I've seen occur in M&A situations.

ComputerWorld summarizes Barclays court motion for relief:

According to the motion, Barclays sent the spreadsheet containing the list of contracts to Cleary Gottlieb at 7:48 p.m. EDT on Sept. 18. The spreadsheet — which contained almost 1,000 rows of data with a total of more than 24,000 individual cells — needed to be reformatted and converted into a PDF file so it could be posted on the bankruptcy court's Web site before midnight. At 11:37 p.m., Cleary Gottlieb sent the converted file to the court, the motion said.

However, contracts that had been marked as "hidden" in the spreadsheet when it was received by the law firm were added to the purchase offer during the reformatting process, according to the motion. Those contracts weren't supposed to be part of the deal; they also were marked with an "N" for "No" in the original version of the spreadsheet, Cleary Gottlieb said in the motion.

PC World points out that:

It's easy for IT people to feel smug when we hear a story like this. Some power user used a fancy feature to pretty up a spreadsheet. Another user made assumptions and gave explicit instructions to a third user, who followed those instructions to the letter.

If only these users had been more tech-savvy -- or had consistent standards for using Excel, or had a better quality-control process, or had taken time to verify the data -- they wouldn't have had this problem. Right?

Except that the spreadsheet was created at one company and sent to a competitor's law firm, so forget about consistent standards. Quality control? This was a one-off spreadsheet for a one-off deal. Time? It didn't exist.

Actually, just spot-checking the finished list against the original spreadsheet would have caught the foul-up: 22 of the 70 items on the first page weren't supposed to be there. (Yes, I counted them myself.)

In other words, being a lot less tech-savvy -- or at least a lot less trusting of technology -- could have saved the day.

It will be interesting to hear what the court decides. In past cases I've seen, courts tend to take the 'caveat emptor' approach to such matters.

I know that many professionals (and software vendors) have given up the ghost in trying to fight against spreadsheet ubiquity. However, if ever there was a requirement for specialist software or rigor in due diligence, this is surely one of those cases. Expect to see more of this type of thing occurring.

Hat tip to @vendorprisey - my old sparring partner at SAP, now at Gartner.