Some of the most valuable components of a business process are the ones that appear to accomplish very little. A backup server sits idle. A validation step slows deployment. A rollback capability requires continuing investment even when nobody uses it. To an efficiency-minded executive, these can look like expenditures awaiting a sufficiently enthusiastic cost-reduction initiative.
Of course, their economic contribution becomes considerably less obscure when everything else stops working.
Microsoft’s massive November 12–13 outage offers an instructive example. What began as a disruption involving Exchange Online spread into other parts of the Microsoft 365 ecosystem, interfering with Outlook, Teams, Copilot, SharePoint-related functions and other services on which hundreds of millions of people and countless business processes depend.
According to CIO.com’s reporting, Microsoft found that a large number of servers had become unexpectedly unavailable, overburdening the remaining infrastructure. A process responsible for establishing secure connections became overwhelmed, while some servers returning to service became unresponsive again once traffic resumed. Engineers investigated configuration changes, redistributed traffic, added capacity and worked to restore stability.
The definitive root-cause analysis was still forthcoming as this article was being written. Consequently, it would be premature to conclude that Microsoft failed to invest sufficiently in redundancy, testing or recovery. Nonetheless, the reported sequence provides a useful case for examining how organizations evaluate the economics of processes that prevent failures—or fail to contain them.
And there is a substantial difference between knowing what it costs to keep a system running and understanding the economic consequences when it doesn’t.
The price of a failure is seldom confined to the organization that caused it
The direct financial consequences to Microsoft could include customer service credits, remediation costs, contractual disputes, legal expenses and eventual reputational damage. But those represent only part of the economic picture.
Banks and trading firms experienced communication disruptions. Shipping companies struggled with coordination. Airlines and government agencies encountered interruptions in routine work. According to Reuters, some organizations resorted to telephones, fax machines and other alternatives. Evidently, the digital transformation revolution has not entirely retired the twentieth century.
Cyber-risk modeling firm Kovrr offered a preliminary estimate of approximately $16 billion in lost productivity worldwide, with potential insured losses around $700 million. The firm characterized these figures as rough estimates dependent on assumptions about the outage’s duration and impact. They’re hardly audited damages, and productivity lost temporarily isn’t necessarily economic output lost permanently.
Nevertheless, given the magnitude, this deserves attention. Even if the ultimate damage proves to be a fraction of such estimates, an infrastructure problem originating inside one technology provider can impose consequences that extend far beyond its own financial statements.
This is an example of an economic externality: costs imposed on parties other than the organization making the underlying investment and operating decisions.
A cloud provider can calculate its infrastructure costs precisely. It may have far less visibility into the financial consequences of an outage at a hospital, an airline, a pharmaceutical company or a government agency dependent on its services. And contractual limitations on liability may mean that its financial exposure is substantially smaller than the economic damage experienced by customers.
Such asymmetry can distort how much an organization considers it worthwhile to spend on resilience.
The economic contribution of what doesn’t happen
In my earlier CIO.com articles, Business transformation needs a true economic approach, not guesswork and Economic process modeling: Business cases beyond cost accounting, I introduced Economic Process Modeling (EPM) as a way to analyze what business processes actually contribute economically, rather than concentrating primarily on their costs and efficiency.
EPM decomposes a process into four component types (information, actions, systems and people) and examines their economic contribution across five dimensions: revenue, cost and friction, risk, option value and information value.
These dimensions are particularly germane to operational resilience because some activities contribute their greatest economic value by preventing undesirable outcomes.
Consider a redundant server that adds $100,000 annually to infrastructure expense. Its direct cost is readily apparent. But suppose it reduces the annual probability of a $10 million business interruption from 2% to 0.5%. The expected annual loss falls from $200,000 to $50,000. That $150,000 reduction in expected risk exposure exceeds the server’s operating expense, yielding a $50,000 net annual economic benefit.
These are illustrative figures, not estimates of Microsoft’s situation. Their significance lies in the calculation, which is straightforward enough for a spreadsheet–or better yet, an EPM-capable modeling tool. The harder part is recognizing that an apparently underutilized resource deserves to have its risk-mitigation contribution measured in the first place.
An ostensibly idle server may therefore be performing some of the most economically valuable work in the organization. Much like a fire extinguisher, its utilization rate is a rather poor indicator of its worth.
Putting the economics into the process
Microsoft’s outage suggests several places where an EPM analysis could inform resilience decisions, both for technology providers and the enterprises dependent on them.
| Process component | What conventional analysis sees | What EPM would additionally examine |
|---|---|---|
| Spare infrastructure capacity | Unused resources and carrying costs | Expected disruption losses avoided under peak and failure conditions |
| Predeployment validation | Testing costs and delayed releases | Financial exposure avoided by detecting defective configurations before deployment |
| Server health checks | Monitoring overhead and processing cost | Economic consequences of incorrect health decisions and cascading capacity loss |
| Traffic management | Throughput, utilization and latency | Downstream losses from overload, retries and correlated failures |
| Rollback and recovery | Maintenance expense and recovery-time targets | Expected reduction in outage losses and the value of preserving recovery alternatives |
| Operational telemetry | Storage and monitoring expense | Value of earlier detection, faster diagnosis and more effective recovery decisions |
Table: Applying EPM’s economic lenses to infrastructure resilience. These are prospective analytical applications, not findings that Microsoft lacked the safeguards described.
A distinguishing feature of EPM is that it can account for economic contributions traditional operating metrics and traditional process modeling may overlook.
Take capacity management. A conventional dashboard might show that a server pool is operating at 80% utilization, suggesting an opportunity to consolidate resources. But what happens when a configuration problem removes a quarter of the available capacity? The surviving servers would be asked to handle more work than they can sustain, potentially producing further failures, retries and escalating demand.
The relevant economic question concerns how much risk exposure that reserve capacity mitigates and whether its cost is justified.
Likewise, a rollback process that is rarely invoked may preserve considerable option value—the ability to pursue changes more confidently because a viable recovery path exists. Diagnostic telemetry may have information value far beyond its storage cost if it helps engineers identify the relationship between an infrastructure change, failing health checks and subsequent overload before the problem metastasizes.
Reliability engineers already understand capacity planning, fault isolation, redundancy and rollback. EPM doesn’t supplant these established disciplines. It provides an economic basis for comparing their contributions, allocating investment and deciding which safeguards deserve protection when budgets tighten.
From process maps to economic shock absorbers
In another recent CIO.com article, The next digital twin isn’t of a machine, it’s of your business, I discussed enriching organizational digital twins with the economics of the activities, information, systems and people they represent.
Operational resilience offers a particularly interesting application of this idea.
Imagine modeling an enterprise’s order-to-cash process, including the cloud services on which it depends. A disruption to email might initially appear to be a communications inconvenience. Yet follow the dependencies and the implications become more consequential: customer orders awaiting confirmation, contracts requiring authorization, invoices delayed, shipments held or exceptions going unresolved because the responsible people cannot coordinate.
A traditional process map can show these dependencies. Process mining and observability tools can reveal where the work stops. Risk analysis can identify where vulnerabilities lie. What an economic model adds is the financial consequence of each interruption and the value of each safeguard that limits it.
That is the real lesson of an outage like Microsoft’s. Resilience is easy to undervalue when it works, because its payoff is a disaster that never appears on the books. Organizations that measure the economic contribution of prevention, not just its cost, are better placed to protect the safeguards that matter when the next disruption arrives.