The Most Expensive Work Is Work You Don't Finish.

Work that's 80% finished delivers 0% of its value — and it keeps costing you the entire time it sits there.

Most portfolio reviews are full of work that's mostly done: the migration that's 70% complete, the feature that's in final QA, the initiative that's "almost there." It reads as progress. It's actually inventory — capital that's been spent, sitting on a shelf, earning nothing until it ships, and it has been quietly costing the organization money the entire time it's been sitting there.

Unfinished Work Is Inventory

Taiichi Ohno, the engineer who built the Toyota Production System, identified work-in-process as one of the core categories of waste (muda) in a manufacturing system — alongside waiting, overproduction, and defects. His insight wasn't that work-in-process is inherently bad; it's that it doesn't create value until it becomes a finished good in a customer's hands, and everything sitting between "started" and "delivered" is a cost the system is absorbing without any return. Delivery organizations tend not to think of half-built features or in-flight initiatives the same way they think of a warehouse full of unsold parts — but the underlying economics are the same.

Inventory Has a Carrying Cost — So Does Unfinished Work

In physical supply chains, this cost is well quantified. Industry benchmarks from supply-chain associations (ASCM/APICS) and the Council of Supply Chain Management Professionals put the all-in annual cost of carrying inventory at roughly 20–30% of its value — capital tied up, storage, insurance, and the risk that it becomes obsolete before it's used. Unfinished delivery work carries a direct analogue to every one of those components: capital already spent on people's time, the ongoing cost of context being held in people's heads instead of shipped, and the very real risk that requirements or market conditions shift while the work sits, making it stale by the time anyone returns to it.

The Cost You Don't See Is Still a Cost

Physical inventory shows up on a balance sheet, so someone eventually has to account for it. Unfinished delivery work doesn't — there's no line item for "half-built features currently depreciating." Don Reinertsen's concept of Cost of Delay, developed to quantify exactly this kind of invisible cost, defines it as the dollar impact of time on the outcomes an organization is trying to achieve. Without putting a number on it, queued and unfinished work is effectively invisible to the people deciding what to fund next — which is precisely why it keeps accumulating instead of getting finished.

What the Data Shows About WIP and Speed

This isn't just a manufacturing analogy transplanted onto software. A 2018 empirical study presented at the ACM/IEEE International Symposium on Empirical Software Engineering and Measurement analyzed more than 8,000 work items completed by five teams over four years at a single software company. The researcher, Dag Sjøberg, found a direct relationship: higher work-in-progress correlated with longer lead times — the more unfinished items a team was carrying simultaneously, the longer each individual item took to actually get done, independent of how much total work the team was capable of completing.

Why Organizations Keep Funding Work That Should Stop

If the economics are this clear, why does unfinished work keep piling up instead of getting killed or finished? Part of the answer is the sunk-cost effect, documented by Hal Arkes and Catherine Blumer in a 1985 study published in *Organizational Behavior and Human Decision Processes*. They found that people are more likely to continue an initiative simply because resources have already been invested in it, even when those resources are unrecoverable and continuing is no longer the rational choice. Arkes and Blumer traced this partly to a desire not to appear wasteful — killing a 70%-funded initiative feels like admitting the first 70% was a mistake, even when the honest economic question is only ever about the value of finishing the remaining 30%.

Finishing Is a Different Discipline Than Starting

Most delivery organizations are structurally set up to reward starting, not finishing. A kickoff is visible, gets a name, and shows up in a steering committee deck. Finishing the unglamorous last stretch of an existing initiative competes for attention against the next shiny kickoff — and loses, because starting something new looks like momentum in a way that quietly closing out old work never does. The result is a portfolio that accumulates open work faster than it retires it, which, per the Sjøberg findings above, makes every single item in that portfolio slower to complete.

The Better Question

None of this argues for finishing everything regardless of merit — some in-flight work genuinely should be killed once its economics no longer justify it. The argument is narrower: unfinished work isn't a neutral, no-cost state while an organization decides what to do with it. It's actively costing capital, clarity, and speed every day it stays open, exactly the way physical inventory does — and that cost belongs in the decision, not outside it.