Paying Twice for the Same Requirements

A global IT organization found that its delivery teams were working out requirements the business had already thought through, and skipping test planning because the picture was still unclear. Making the business own acceptance criteria, with leadership holding the line, stopped the company paying for the same work twice.

02 SCOPE
The condition

Delivery teams were skipping a required quality step, and everyone assumed they were simply too busy.

A Quality Gap Everyone Could Explain

The survey results came back during a working session with the PMO leader of a global IT organization. One finding stood out. A meaningful share of development teams said they did not write a test plan before starting a new project. The organization's own quality assurance process required one, and it was in the middle of an external process maturity audit. A gap like that was exactly what auditors would notice, and she would be the one asked to explain it.

The room settled on the obvious answer almost immediately. Capacity. The teams were overloaded, and test planning was one of the first things to disappear when delivery pressure rose. We told them that had been our hypothesis too, and that the data said otherwise.

The Cause Sat Upstream

The survey had also asked teams why they skipped test planning. Resource constraints showed up in the answers, but they were not the most common reason. The most common reason was that requirements were too unclear at the start of a project to make a test plan practical. Teams were waiting for clarification that often arrived only when they were well into delivery. By then the step had been skipped, and nobody went back to it.

The PMO leader saw it before anyone else said it. The quality gap was not a quality gap. It was a scope gap that showed up in quality. Teams could not plan how to test work when nobody had defined what would make it pass. For years the organization had treated unclear requirements as a delivery problem.

Underneath sat a duplication nobody had chosen. The business thought each request through once. Delivery teams then worked the requirements out a second time, later, under a different set of assumptions. Two groups were doing the same work, neither could see the other's version, and neither version was good enough to test against.

Figure 1
Where the test plan disappeared
  1. Business thinks it through The first pass at the requirements.
  2. Where it broke Request arrives unclear No shared definition of success.
  3. Knock-on delay Team waits for answers Clarity arrives mid-delivery.
  4. Paid twice Requirements rebuilt A second pass, new assumptions.
  5. The symptom Test plan skipped Not revisited once work is under way.
The missing test plan was the last link in a chain that started when a request crossed from the business to delivery without a shared definition of done.

Nothing Without Acceptance Criteria

The fix was a single rule. Every work item entering the system would carry acceptance criteria: specific, testable statements of what the work had to do to be considered complete. Delivery teams would not write them. The business requesters would, with delivery teams consulted to make sure each criterion was testable and feasible. Until the criteria existed, the work would not move into delivery.

The point was not to hand the business extra work. It was to make each request realistic and understood by the people who had to fulfill it, and to put the requirements work in one place, owned by one side and checked by the other.

The PMO leader named the risk right away. The business had treated requirements as IT's problem for years and would push back. We agreed it was the hardest part. A rule enforced from inside the PMO alone would not survive. It needed sponsorship from the IT leadership team, visible support from the CIO, and a clear expectation that defining success was a shared responsibility. Teams could only hold the line if leadership gave them cover when business units objected.

Figure 2
Who owned the requirements
Owned by no one
  • Business hands over a request
  • Delivery redoes requirements
  • Test planning gets skipped
  • Delivery absorbs the cost
Owned at the source
  • Business writes the criteria
  • Delivery checks testability
  • No criteria, no delivery
  • Leadership backs the returns
Before, the requirements were everyone's job and no one's. After, the business owned them at the source and delivery checked them before committing.

The Week the Rule Became Real

Over the next six weeks we ran workshops with representatives from every major business unit on how to write acceptance criteria and why they were being asked to. The first session was the hardest. Several people came ready to challenge the policy, and one asked directly why the business was now writing requirements when it paid IT to figure them out.

The PMO leader's answer, which we had prepared with her, was that the company had been paying for this work twice: once when the business thought the request through, and again when delivery teams reconstructed the requirements later under different assumptions. Neither pass was getting the job done well. Writing acceptance criteria together was not more work. It was how the rework would stop.

By the third session, word had spread, and the questions turned practical. What makes a criterion testable? What happens when understanding changes during delivery? In week four, delivery teams began returning work items that arrived without complete criteria. Predictably, a few business leaders complained to the CIO. He held the line, and made clear the expectation applied to everyone, including the leaders who reported to him. Without that, the rule would likely have collapsed at the first sign of resistance.

Two colleagues reading through a document together.

The Audit Finding That Never Came

By week eight, work was arriving with acceptance criteria and teams were building test plans against them. The number of projects entering delivery without a test plan kept falling. Six months later, the next audit cycle arrived. The quality assurance gap that had been forecast as a likely finding was not there. Teams were following the process because the conditions needed to follow it now existed.

At the final review for that workstream, the PMO leader put it plainly. The organization had treated delivery quality as a delivery problem for a long time. It had really been a scope problem that delivery teams were absorbing, and once requirements arrived intact, the rest of the system was already capable of delivering well.

The duplicate effort had never appeared on any report. Nobody decided to pay for requirements twice. It happened because the thinking that defines a request lived on one side of a handoff and the work that depends on it lived on the other, and nothing made the two meet before delivery began.

Figure 3
Eight weeks and one audit cycle
  1. Survey
    What surfaced Test plans often skipped
  2. Same survey
    The real cause Requirements, not capacity The most common reason given.
  3. Weeks 1 to 6
    What changed Workshops with each unit How to write acceptance criteria.
  4. Week 4
    The test Incomplete work returned The CIO backs the teams.
  5. Week 8
    What changed Criteria in, plans follow
  6. Six months on
    What changed Audit gap not there
The rule held because it was tested early and leadership backed it. The audit six months later confirmed the change had stuck.

When two groups each do the same thinking and neither can see the other's, the waste is not only the duplicate effort. Neither version ends up good enough to build from.

Share With Your Team

Choose who it is for. You send it yourself from your own email; Agileon never writes to your colleagues.

Not sure where to start?

Answer five questions about where work stalls. We send back a starting plan: where to begin and what to look at first.