A team finishes its work sooner. The customer still waits.

That gap deserves attention before the organization celebrates a productivity gain.

The work may be ready for review, waiting for another specialist, or blocked by a decision that nobody has time to make. Faster preparation helps the team. Whether it helps the customer depends on what happens next.

For transformation leaders, this is a practical distinction. Local speed and end-to-end performance are different things to measure.

In my previous article, “Your AI pilot worked. Who owns the workflow now?”, I examined ownership after a successful pilot. This edition follows the work across organizational boundaries, where several owners can have legitimate priorities but depend on the same limited capacity.

The October 2026 EY-Parthenon survey of 1,200 CEOs offers relevant context: 23% reported productivity gains being absorbed by complexity, regulation, risk or new work demands.[2] This is self-reported evidence across productivity improvements, not a measured loss caused specifically by AI or coordination.

Follow the work beyond the team

Consider an illustrative example, not a client case. A team improves its preparation process and sends sixteen complete proposals a day to specialists who can review ten. If every proposal needs the same review and capacity remains unchanged, six proposals accumulate each day: thirty over five working days.

The local improvement is real. So is the growing queue.

The next management decision depends on the purpose of that review. Some checks may be essential. Others may be duplicated, applied to work that does not require them, or triggered by incomplete information upstream.

Adding reviewers is one option. Improving the quality of incoming work, changing the release rate or redesigning the review may be better options. Removing the review without understanding its purpose could simply move cost and risk elsewhere.

The point is to find out which condition actually limits the result.

What I mean by the coordination tax

I use “coordination tax” here as a working label for avoidable effort and delay at the boundaries of work: repeated explanations, competing requests, unclear acceptance conditions and decisions that keep returning to the same meeting.

It is not a standardized metric, and it is not an argument against collaboration.

A necessary safety review is valuable work. A discussion that resolves a real trade-off can save substantial rework. Slack capacity can protect service reliability when demand varies.

The useful distinction is between coordination that improves the outcome and coordination that compensates for an unresolved design problem.

Microsoft’s June 2025 Work Trend Index report describes fragmented working days using aggregated Microsoft 365 activity and survey evidence.[1] It provides context for examining coordination, but it does not establish how much of any particular organization’s communication is wasteful.

More messages alone are a poor diagnosis. Follow a real item through the work and establish what each exchange made possible.

Make the boundary visible

Choose one outcome that matters to the business. It might be a customer receiving a usable proposal, a change reaching production safely, or a service request being resolved correctly.

Then examine a small, recent sample of completed and still-open items with the people who handled them.

For each item, record when it became ready for the next step, when that step began, whether it returned for rework, and what decision or information was missing. Include delayed cases rather than selecting only successful deliveries.

This creates a more useful starting point than asking a department to defend its performance.

Keep the conversation concrete:

• Where did this item wait?
• What was needed before work could continue?
• Which record supports that explanation?
• What else could explain the delay?

The evidence may reveal a shared specialist serving several strategic initiatives. It may reveal an unclear request, rather than insufficient capacity. It may also show that the delay was justified.

Each finding calls for a different response.

Treat shared capacity as a portfolio decision

When several initiatives depend on the same scarce capability, separate project plans do not create additional capacity.

Someone with authority across those initiatives needs to decide what enters the shared queue, what can wait and what trade-off the organization accepts. Teams should contribute the operational evidence. Leadership must resolve competing commitments that teams cannot resolve themselves.

Make that trade-off explicit at the level where the competing commitments meet. Identify the business outcome that takes precedence, the work that will wait and the person authorized to resolve exceptions. Give the affected teams a clear decision they can execute.

For example, leadership may protect a specialist’s time for one critical release and defer a lower-priority initiative. That choice has a cost: the deferred initiative loses time or opportunity. Make the consequence visible, explain the choice to those affected and set a review date.

An urgent exception should identify which commitment it displaces. Otherwise, every sponsor can keep a local promise while the organization makes a collective promise it cannot deliver.

The practical implication I draw is to test the whole path before treating local time savings as a portfolio benefit.

Test an improvement without distorting the work

Agree one change with the people doing and receiving the work. Set a review date suited to the volume and risk of that workflow. Two weeks can be a useful initial observation window for frequent, comparable work; it is not a universal evaluation period.

Track elapsed time to the usable outcome, the age of waiting items, rework, quality and workload. Record changes in demand and case complexity. Otherwise, a quieter week can look like a successful intervention.

Do not reward volume at the expense of the result. Stanford’s October 2026 account of workplace research describes paralegals moving into more complex work while handling fewer cases.[3] The comparison was observational, with differences in training and support as well as managerial framing. It challenges simplistic volume measures; it does not prove a universal intervention.

If the queue shrinks but mistakes rise or pressure moves to another team, the result needs reconsideration. If nothing improves, revisit the explanation before expanding the intervention.

A useful transformation decision makes it easier for people to complete work that matters. The test is what reaches the customer, with what quality and at what cost, after the work has crossed all the organizational boundaries.

At the next portfolio review, take one delayed outcome and trace the handoff that held it back. Make the smallest change the evidence supports. Then check whether the outcome improved.

Sources and further reading

1. Microsoft WorkLab, “Breaking down the infinite workday”, 17 June 2025. https://www.microsoft.com/en-us/worklab/work-trend-index/breaking-down-infinite-workday

2. EY-Parthenon, “How can CEOs turn AI productivity gains into lasting value?”, 1 October 2026. https://www.ey.com/en_gl/ceo/ceo-outlook-global-report

3. Stanford HAI (Human-Centered Artificial Intelligence), “Want Employees to Embrace AI? Stop Selling It as a Productivity Tool”, 7 October 2026. https://hai.stanford.edu/news/want-employees-to-embrace-ai-stop-selling-it-as-a-productivity-tool

Between the Boxes | Milo Izzinosa | Enterprise Transformation & Execution Lead

If a shared dependency is holding back a transformation priority, I am available to discuss the execution challenge and a practical next step.

Originally published on LinkedIn on 8 October 2026 as part of Between the Boxes. Read the original LinkedIn article.

Explore our approach to transformation and execution, or contact eXcelleRes to discuss an execution challenge.