There is a particular meeting that anyone who has worked around large programmes will recognise. The pack says green. The workstream leads say green. And somewhere in the appendix, a milestone that was due in March is now shown against June, with a note explaining that the sequence has been optimised.
Nobody in that room is lying. That is the part worth understanding, because if the problem were dishonesty the remedy would be simple. It is not. The remedy is structural, and it has to be built before the programme starts rather than discovered halfway through.
Why status drifts upward
Status is assembled from the bottom. A team lead knows a task is late but also believes it is recoverable, so reports amber with a plan. The workstream lead receives four ambers, each with a plan, and reports green with risks noted — because a workstream is not amber merely for containing difficulty. The programme office aggregates six workstreams reported green and produces a green pack. No step in that chain is unreasonable. The aggregate is nonetheless false.
Layer on top the ordinary human calculus. Reporting red invites scrutiny, and scrutiny consumes the time needed to recover. A lead who believes recovery is possible has a rational incentive to defer the bad news by one cycle. It is usually one cycle, and then one more.
Optimism is not the failure. Aggregation is. Each layer rounds towards confidence, and six layers of rounding produce a number with no relationship to the ground.
The dates that move quietly
The second mechanism is subtler. A milestone slips, so the plan is re-sequenced. Downstream work is brought forward, the critical path is redrawn, and the end date holds. This is presented, correctly, as good management. What is rarely stated is that the float has been consumed. The programme is now more fragile than it was, and no status colour has changed to say so.
By the time float is exhausted, the next slip transmits directly to the end date, and it does so without warning — because every previous slip was absorbed silently. The transition from green to red is therefore almost never gradual. It is a step change, and it arrives at the worst possible moment.
What actually helps
Four things, none of them exotic.
Define completion unambiguously. A milestone that is met when a document is submitted is not a milestone; it is an activity. Completion criteria should be objective enough that two people looking at the same evidence reach the same answer. Most milestone disputes are definitional, not factual.
Report float, not just dates. If the pack showed remaining float on the critical path alongside the end date, the erosion would be visible from the second month. A programme whose float has halved is telling you something even while its dates hold.
Separate the reporting line from the delivery line. Where the party being measured also produces the measurement, drift is inevitable regardless of anyone's integrity. An owner needs its own instrumentation — not because the integrator is untrustworthy, but because that is simply not the integrator's office.
Make amber survivable. If amber triggers an inquisition, nobody uses it. If amber triggers help — a decision unblocked, a resource freed, a dependency resolved — it gets used early and does its job. Governance shapes reporting behaviour far more than any instruction to report honestly.
The uncomfortable first week
When an independent baseline is established on a programme already in flight, the first report is usually worse than the last one produced internally. That is not evidence that the reviewer is harsh or the team was hiding something. It is what happens when six layers of rounding are removed at once.
Sponsors who understand that in advance handle the moment well. Those who do not tend to shoot the messenger, revert to the comfortable number, and meet the same problem again nine months later with less money and fewer options.
Seshail Kamanna advises boards and CXOs on technology programmes, with nineteen years across twenty-four sectors. More on programme management offices or get in touch.