Our earlier brief on decision risk looked at spreadsheets that quietly drive day-to-day operational decisions, unglamorous, easy to overlook, and easy to underestimate. This one looks higher up the chain: two cases where a spreadsheet error reached the board level itself, informing acquisitions worth billions of dollars, reviewed by name-brand advisors, and still not caught before the decision was made.
Neither case required a regulatory failure to happen. Both required only one thing: a piece of spreadsheet logic that nobody outside the person who built it had independently checked.
Seniority does not substitute for verification. A board is only as well-informed as the numbers nobody in the room actually double-checked.
Two boards. Two spreadsheets. Two decisions made on numbers nobody had verified.
Both of the cases below involve sophisticated, experienced institutions, a major investment bank in one, a leading global advisory firm in the other. Neither lacked expertise. Both trusted a spreadsheet that had never been independently reviewed by anyone outside the team that built it.
In the chaos of Lehman Brothers' 2008 collapse, Barclays moved to acquire a portion of its assets. A spreadsheet listing the contracts Barclays intended to purchase contained additional rows marked to be excluded, hidden, not deleted. When the file was reformatted and converted to PDF for the bankruptcy court filing, those hidden rows became visible content in the official document. Barclays found itself legally bound to 179 contracts it had never intended to acquire, discovered only after the filing deadline had passed.
Lazard, advising SolarCity's Board on its sale to Tesla, presented a valuation range to the Special Committee based on a discounted cash flow analysis. Weeks later, a "computational error" was discovered: the underlying spreadsheet had double-counted roughly $400 million of the company's projected debt, understating SolarCity's equity value by that amount in the very range the Special Committee had relied on to assess the deal.
"A hidden row and a double-counted formula are the same failure, wearing different clothes. Both are logic nobody outside the builder ever independently checked."
Why expertise didn't catch it
It would be easy to read both cases as carelessness. They were not. Barclays' error was discovered only because of an unrelated file-conversion step, reformatting for a PDF, that happened to expose what had been deliberately hidden. Lazard's error was a "computational error" inside a discounted cash flow model, precisely the kind of formula logic that looks correct at a glance and only reveals itself under deliberate, independent scrutiny.
Both errors were genuinely invisible to a normal read-through. That is the actual point: reading a spreadsheet is not the same as validating it. Neither Barclays' lawyers nor SolarCity's Special Committee were negligent for not catching these errors by eye. The failure was structural, nobody's role in either process was to independently trace and stress-test the underlying logic before it informed a board-level decision.
As with our first decision-risk brief, this one deliberately avoids anchoring to a specific regulation. Neither Barclays' nor Lazard's error was a compliance failure in the formal sense, both were caught, eventually, and neither institution was found to have broken a rule. The risk here is not regulatory. It is structural: even the most sophisticated institutions can carry spreadsheet logic all the way to a board-level decision without anyone independently validating it first.
For organisations without a Board Special Committee or a bankruptcy court's document conversion process to accidentally surface an error, that structural gap is worth taking seriously on its own terms.
Independent review, before the number reaches the table.
Both cases above share a common thread with the EUC risk we help Financial Services firms close every day: a spreadsheet's logic, its hidden rows, its formula structure, its assumptions, was never independently reviewed by anyone outside the person who built it. A Greywood Analytics deconstruction engagement applies exactly that independent scrutiny, for each asset in scope, we deliver:
- Plain English Summary — what the spreadsheet actually calculates, in language a board member could verify against their own understanding.
- Dependency Map — every formula chain and input traced explicitly, including anything hidden, nested, or easy to miss on a normal read-through.
- Dependency Mapping Narrative — the written analysis of exactly how the logic works, and where it could silently fail.
- Risk & Assumption Narrative — every embedded assumption tested, not assumed correct because it looked reasonable.
- Risk Register — a clear, prioritised view of what needs independent verification before it informs your next major decision.
The goal is simple: by the time a number reaches a board table, someone other than its builder should already have checked it.
First deliverable pack typically received within 2–3 working days of engagement start.
The question worth asking before the next major decision
Both Barclays and Lazard eventually caught their errors, one by chance, through an unrelated file conversion, the other closer to the deadline, through the diligence of a Special Committee's review. Neither institution had a process specifically designed to catch it. They got lucky, and luck is not a control.
The question worth asking before your organisation's next major, spreadsheet-informed decision is not "has anyone read this." It's "has anyone independently traced and tested the logic, separately from the person who built it." For most organisations, the honest answer is no, not because anyone was careless, but because nobody's role was ever explicitly to do it.
"Being read is not the same as being checked. Both Barclays and Lazard had readers. Neither had, by design, an independent check."Greywood Analytics
If a major decision at your organisation is resting on spreadsheet logic that's never been independently traced and tested, we should have a conversation. Every engagement begins with a no-commitment scoping discussion.
Start a Conversation