The True Cost of a Failed Rollout

The failed IT rollout's cost never fits in its budget line. The direct overrun, the number that reaches the board, is usually the smallest entry on the real invoice. The full accounting spreads across the business for years, which is why deployment discipline is cheap at almost any price. Here's the itemised version.

The visible costs

The overrun itself. Revisits, extended crews, emergency freight, contract disputes: a troubled program routinely lands at 150 to 200 per cent of budget. Painful, quantified, and the least of it.

Trading losses. Stores that opened late, lanes that couldn't sell, branches that turned customers away. These losses have the awkward property of being real money that appears on no project ledger, so they're systematically undercounted, and store managers remember them long after finance stops looking.

The submerged costs

The support debt. A rollout that limps to "done" at 90 per cent consistency bequeaths an estate of exceptions: sites on workarounds, undocumented variances, an asset register that's partly fiction. Every support interaction for the next five years pays a small tax to that inconsistency, the snowflake problem, installed at scale, deliberately paid for.

The stalled roadmap. Whatever the rollout was meant to enable, new payments, new platform, new formats, waits. The business case's benefits column starts a year late while its costs column ran over, a double hit that rarely gets reunited in one report.

The organisational scar tissue. The subtlest and longest-lasting entry. A failed program teaches an organisation to fear programs: the next initiative gets scoped smaller, approved slower and staffed more defensively. Competitors who deploy confidently compound their advantage every cycle while the burned organisation relearns trust. Ask anyone who has tried to get a second rollout approved after a first one failed.

Where the money actually went

Post-mortems on failed rollouts converge on the same short list: no real site audits, no staging, no piloted playbook, thin governance, and a partner selected on price and adjectives. Which is to say, the seven challenges, unaddressed, on schedule. Failure is rarely exotic. It's the standard risks, undiscounted.

The comparison that matters

Weigh the full invoice, overrun, trading losses, support debt, stalled benefits, scar tissue, against the cost of doing it properly: audits, staging, pilots, real governance, an evidenced partner. The disciplined version costs single-digit percentages more upfront and deletes a failure mode measured in multiples. Framed as insurance, it's the cheapest policy in enterprise IT; framed as "extra process", it gets cut, and the invoice above gets paid instead.

The 673-store program that finished three weeks early and the rollout that stalls at 90 per cent forever run on the same hardware. The difference was never the technology.

Prefer the first kind of program? Speak to an expert.