By the time delivery confidence is publicly questioned in steering, the cost of correction has already multiplied. The signal worth reading is earlier.
The pattern
Programme drift rarely arrives as a single failure. It accumulates as small concessions: a deadline pushed by a week, a status that shifts from amber to amber-green without new evidence, a vendor SLA that everyone knows is informal but no one has the authority to renegotiate. By the time RAG turns red, three to five months of corrective leverage have already been lost.
Senior buyers do not pay for the visible failure. They pay for the silent compounding that preceded it.
Three signals to read in week six, not week twenty
Three operational patterns reliably precede public RAG drift by one to three quarters. They do not require new instrumentation. They show up in the same artefacts steering already produces.
Signal one: the narrative-detail gap.
The programme summary in steering is increasingly optimistic. The detail underneath, when probed, cannot be defended in the same language. Decisions taken at steering are summarised but not documented with owners. This is the earliest signal and the most reliable.
Signal two: vendor SLA reporting that has gone procedural.
Quarterly reviews still happen. They include the same slides, the same RAG colours, and they are read silently. Two or more vendors have missed commitments without escalation. No one has documented what would constitute grounds for replacement.
Signal three: cross-market divergence without coordination.
Two or more markets in a multi-country programme are solving the same operating problem in different ways. Headquarters has stopped asking why. This is the structural signal that operating discipline has eroded below the executive layer.
What changes after intervention
The work is rarely heroic. It is rebuilding the artefacts that already existed but had stopped functioning: a steering pack that surfaces decision points in advance, a vendor scorecard that triggers escalation rather than reporting it, a cross-market governance cadence that produces shared decisions instead of parallel ones. Sixty days is enough to install the discipline. Six months is enough to make it survive handover.
The senior-advisory case
Programme drift is rarely solved by adding more meetings or more reporting. It is solved by installing one operator with the seniority to make decisions that the existing team has been deferring. Most of those decisions were already obvious internally. They needed an external mandate to be made.
Read the signals at week six. Acting at week twenty costs three months of leverage.