The dashboard is green. The customer is still waiting.
Tickets have closed, the quarterly number has been reached and somebody has prepared a presentation about operational excellence. Downstairs, the same work is being reopened, corrected or passed to another team. The reporting period is over. The problem is not.
This is the Digital Kolhoz: an organisation that learns to satisfy the plan while losing sight of the work the plan was supposed to improve.
The Soviet comparison is deliberate. It is an analogy about incentives, not a claim that a company and a state are equivalent, or that one management practice explains the Soviet Union’s collapse. The useful lesson is narrower and more uncomfortable: a reporting system can reward behaviour that damages the operation it claims to control.
The number becomes the product
Consider three hypothetical instructions.
Close more support tickets. A team can make the count rise by closing difficult requests before they are resolved. Reduce the bug backlog. A manager can reclassify defects instead of fixing them. Increase completed claims per day. A department can favour easy claims while complicated ones grow older in the queue.
None of those outcomes is inevitable. That is precisely why the count alone is inadequate. It cannot distinguish useful improvement from a change in how work is selected, recorded or labelled.
A target becomes dangerous when the organisation rewards the visible number and stops examining what produced it. The worker who solves a difficult problem may look slower than the worker who passes it on. The team that records defects honestly may look worse than the team that conceals them.
You can buy a better dashboard. It will display the same mistake more beautifully.
Punish an improvement, then wonder why nobody improves
There is another way to teach people to hide capacity: turn every exceptional result into the next compulsory minimum.
The incentive problem is well established. In their 1983 study of the Soviet enterprise, Michael Keren, Jeffrey Miller and James R. Thornton analysed how the ratchet affected managers’ output choices. Their model did not find that withholding capacity was inevitable in every situation. It did show how an incentive arrangement could make producing at capacity unattractive.
Imagine a team that clears a backlog through an exceptional effort. Management raises its permanent quota without investigating the overtime, postponed maintenance or unusually simple work behind the result. The lesson is immediate: revealing what you can do today creates a liability tomorrow.
Then leadership asks why estimates contain buffers and why improvement proposals arrive slowly.
Before blaming the people, examine the bargain you have offered them. If every improvement becomes a demand for more, you have given them a reason to conceal improvements.
The end-of-period storm
Soviet planning also gives us a name for frantic work around reporting deadlines: shturmovshchina, usually translated as “storming.”
This is more than a colourful factory anecdote. Vahe Lskavyan’s 2022 study examined declassified records of KGB agent recruitment in Latvia from 1961 to 1990. It found recurring mid-year and year-end recruitment surges associated with reporting periods. The study connected these patterns to quota incentives and examined differences across recruitment groups and areas.
The study concerns a particular institution and dataset. It does not prove why your team closes tickets on Friday. It does give a documented example of reported activity gathering around the calendar that governs performance assessment.
In a modern organisation, investigate the equivalent pattern: a quiet reporting period followed by a sudden wave of completions. Were the underlying tasks finished then? Were records updated in a batch? Did a release window determine the timing? Were checks rushed to reach a number?
Those explanations require different responses. A spike is a reason to investigate, not a verdict on someone’s intentions.
When a team sacrifices quality to satisfy the reporting calendar, calling the ritual “Agile” does not make it so. That is storming with a different meeting invitation.
Make the target answer to the work
Start with the records behind the result. Define what “completed” means. Check whether completed work is later reopened. Separate elapsed time from recorded processing time. Examine the work left waiting as well as the work that crossed the finish line.
Keep the original sequence. A total can hide a final-day rush. An average can hide a growing tail of old requests. A measure of variability cannot explain why the variation happened.
Then examine the management decision. What behaviour does the target reward? What can improve the score without improving the service? Who pays for the work displaced into the next period or another department?
At Ghost Citadel, we want detailed process knowledge in people’s hands so they can ask those questions. A metric should help you examine an operation. It should never become an excuse to stop examining it.
The test is practical. When the number improves, can you show what improved for the people relying on the work?
If the answer is another dashboard, keep looking.
References
- Keren, M., Miller, J. and Thornton, J. R. (1983). The Ratchet: A dynamic managerial incentive model of the Soviet enterprise. Journal of Comparative Economics, 7(4), 347–367.
- Lskavyan, V. (2022). Storming in agent recruitment: Evidence from declassified Soviet secret service files. Journal of Economic Behavior & Organization, 200, 973–990.