The annual planning cycle prefers a number. A development programme may instead have a promising observation, a set of conditions under which it was made and an unresolved question about how widely it applies. The pressure to bridge that gap is understandable. If the company is spending money on an improvement, colleagues want to know when the benefit can enter the plan. The difficult leadership task is to give them a useful answer without assigning the evidence a meaning it has not earned.
Our wind-control investigation put that task in front of me. The first phase suggested that a selected control approach deserved more work in one direction sector, while another sector remained inconclusive within the engineering team's stated uncertainty. I supported an extension and declined to insert an annual production gain into the portfolio plan. Those choices can look inconsistent if development expenditure is assumed to require an immediately quantified operating benefit. They are consistent when the next investment is explicitly an investment in learning.
The distinction should be agreed before a trial begins. A programme intended to investigate a mechanism, a campaign intended to estimate an annual effect and a deployment intended to deliver a service are different commitments. They may be connected, but one does not automatically become the next because its initial observation is encouraging. I want the sponsor to know what decision the first phase can support and what would remain necessary afterward.
Measure the group whose performance you want to change
Wake steering asks a different question from maximising the output of each turbine independently. Changing how an upstream machine faces the wind can alter the conditions experienced by machines behind it. A local loss may be compatible with a group-level gain. That means an assessment focused only on the upstream machine could miss the intended effect, while a presentation focused only on the most improved downstream machine could exaggerate it.
I asked the team to define the relevant group and the comparison before the measurement campaign. That did not make the engineering simple. Wind conditions vary, and the behaviour of the machines and the surrounding flow has to be examined by specialists. It did make the management question clearer: we were considering whether a control approach improved the account of the selected group under the conditions being investigated. We were not collecting isolated favourable machine readings for a portfolio presentation.
The operational account also mattered. A production observation does not by itself settle turbine loads, equipment limitations or the permissions required to operate a control approach. Maren Solbrecht kept those questions in the review. I supported that wider examination without claiming that the first phase had demonstrated an adverse or favourable load outcome. It would have been as misleading to invent a problem as to declare those issues resolved because the energy observation looked promising.
The point for a board is not to learn the detailed control settings. It is to understand whether the measured quantity corresponds to the claimed benefit and whether important consequences remain outside the result. That is a question directors can ask across many technologies. The answer needs enough technical substance to be credible and enough clarity to show how it changes the next commitment.
Keep the eligible observations and the missing conditions visible
Our first phase planned eighty observation blocks across two selected wind-direction sectors. The team set eligibility rules in advance, including operating availability and the wind conditions needed for a usable comparison. Sixty-two blocks met those rules. The remaining eighteen stayed in the record with their exclusion reasons. Keeping that account visible mattered because the usable observations were not a complete description of every condition encountered during the campaign.
Alternating the control condition helped the comparison, but it did not erase atmospheric differences. Nor did two selected sectors represent a full annual distribution of wind conditions. I resisted the temptation to turn the encouraging sector into a general percentage and multiply it across the year. The calculation would have looked precise while resting on a transfer that the programme had not justified. A larger spreadsheet would not have repaired the missing evidence.
The inconclusive sector also deserved an accurate description. It did not establish that the approach had no effect there, and it did not give us permission to assume the favourable result from elsewhere. The next investigation needed to improve our understanding of those conditions. This is where a technology sponsor has to explain uncertainty without using it as a vague reason for endless study. The extension should specify the questions it will examine and the decisions that those answers could change.
At the July review, the five-month first phase had cost €870,000. I authorised a €390,000, four-month extension covering additional conditions and specialist review of the load and operating evidence. At September 2026 that work was still under way. The account therefore contains a completed first phase and an ongoing extension. I do not describe the second phase as a success in advance or present the combined expenditure as a realised production improvement.
Give the planning cycle a decision it can actually use
Declining an annual benefit did not mean giving the planning team no answer. I could state the authorised development cost, the expected review point and the conditions under which a later recommendation would be considered. I could explain why the existing operating plan retained its prior production basis. Those were usable decisions. They were less attractive than an immediate uplift, but they did not require the finance team to conceal a technical assumption inside the total.
I also owed the commercial sponsor an explanation of why continuation was proportionate. The promising sector supported a reason to investigate further; it did not settle the value of a fleet deployment. The extension's cost and scope could be compared with the potential relevance of the technology and with other development priorities. That comparison should remain open to challenge. A programme does not become entitled to further funding merely because it has generated interesting data.
This experience has changed the question I ask at the start of a trial. Alongside the engineering protocol, I want an agreed description of the permitted planning use of the results. Who may quote a benefit, at what boundary and for which conditions? What would have to be demonstrated before the figure enters an annual plan? Answering those questions early reduces the pressure to settle them after a favourable observation has acquired a sponsor.
A thoughtful technology account can therefore contain enthusiasm and restraint at the same time. I can believe that an approach deserves more attention while acknowledging that its annual effect remains unproven. The value of leadership is in making that position actionable: a bounded investigation, an honest operating plan and a clear next review. The organisation can then continue learning without borrowing certainty from a future programme that has not yet been completed.