What Makes a Gate Review a Real Gate
A gate review is only a gate if a design can genuinely fail to pass through it. Most so-called gate reviews are not gates at all. They are scheduled checkpoints a design walks through regardless of what the review actually concludes.
A gate exists to stop a design from proceeding to the next phase until it has demonstrably satisfied the conditions of the current one. This connects directly to a simple standard already worth applying at any review, only items that are genuinely closed should be allowed through. Anything still under review, or held pending an unresolved condition, has no business passing a gate that is meant to actually gate something.
The actual standard
Each phase gate requires full satisfaction of that phase’s conditions, not a partial or mostly complete pass. This is a deliberately high bar, and it needs to be, because a phase gate that accepts anything less has already started letting gaps through that will not simply disappear.
What happens when the standard isn’t met
If full satisfaction cannot be demonstrated at a gate, there are two legitimate paths forward, and only two. Either the outstanding work is genuinely completed and the design comes back to the gate properly, or the gap is significant enough that a formal rebaseline discussion needs to be ordered, acknowledging that the original plan or schedule itself may need to change. What is not a legitimate path is proceeding anyway while noting the gaps for later attention.
Why partial passes are dangerous
Allowing a design through a gate at anything less than full satisfaction does not make the gap disappear. It carries the gap forward into the next phase, and because each phase builds directly on the one before it, an accumulation of small partial passes compounds. By the time the full system is assembled, what started as a minor incompleteness at one gate has become a genuine nonconformity or deviation from the design specification, at a point far harder and more expensive to trace back to where it actually originated.
Why most gate reviews aren’t real gates
In practice, the programme schedule has usually already committed downstream work to a fixed start date before the gate review even happens. The review takes place, but the design proceeds regardless of what the review actually concludes, because stopping it would mean admitting the schedule itself was wrong, which is a harder conversation than simply letting the design through and hoping the gaps resolve themselves later.
How to tell a real gate from a scheduled checkpoint
There is a simple test. Ask what would genuinely happen if the review concluded the design should not proceed. If the honest answer is that it would proceed anyway, regardless of that conclusion, it was never a gate. It was a checkpoint on the calendar, dressed in the language of governance.
Why a gate needs to live in the culture, not just the process
A gate that depends entirely on one person actively enforcing it in the moment is fragile, resting on that person’s vigilance every single time it matters. A gate that is genuinely embedded into the culture is different. Engineers themselves come to treat incomplete work as unfit to present, not because a reviewer will catch it, but because presenting it that way was never something the team considered acceptable in the first place. A gate guarded from outside can be worn down by pressure. A gate embedded into culture is far harder to erode, because it does not rely on a single point of enforcement.
Why gates should be benchmarks, not milestones
A milestone is a point in time, a date on a schedule marked as reached regardless of the actual condition of the work behind it. A benchmark is a standard of completeness that genuinely has to be met. Treating a gate as a milestone quietly imports the idea that hitting the date matters more than satisfying the standard, which is precisely the failure this entire discipline exists to prevent.
What a gate’s actual output should be
A gate review should not simply conclude passed or failed as an isolated verdict. Its real output is the specific next action required, proceeding, because the standard was genuinely met, completing outstanding work and reapplying, because it was not, or initiating a rebaseline discussion, because the gap is significant enough to require it. A gate measured only against whether a target date was hit has already lost sight of what it was built to protect.