Reporting that only travels upward is not reporting
A report that is summarized at every level, by people accountable for what it contains, arrives at the top as a description of the reporting process rather than of the project.
Capital project reporting is built as a hierarchy. Site reports to package, package to project, project to program, program to sponsor, sponsor to board. Each level summarizes the level below.
This is a sensible way to compress information and a poor way to detect a problem, because each summarizing step is performed by someone accountable for the content being summarized. Nobody has to act dishonestly for the result to be systematically optimistic.
The mechanism, without anyone lying
At each level the author faces the same practical question: which of the things that are currently uncertain are worth escalating.
Escalating an issue invites attention, requests for analysis, and the implication that it is not under control. Not escalating it is defensible if there is a plausible route to resolving it before the next report — and there usually is, because most issues are resolved at the level where they arise. That is what management is.
The consequence is that each level filters out the problems it believes it can handle. The filter is individually reasonable and cumulatively severe. By the fourth summarization, what reaches the board is the set of problems that four successive managers were unable to resolve and unwilling to conceal.
Nothing in the chain is false. What is missing is everything that was still recoverable at the moment it was left out, which is exactly the population an owner would want to know about.
What the top of the chain receives
The characteristic output is a report that is accurate in every particular and wrong in aggregate. Status is amber, trending green. The schedule is described as challenging but achievable. The change position is under review. Risks are listed with mitigation actions in progress.
None of that is a lie, and none of it can be tested from within the document. The report describes conclusions without the evidence that produced them, so a reader can only accept or reject the conclusions on trust.
The failure becomes visible in one of two ways. Either a problem finally clears every filter simultaneously, which presents as a sudden deterioration from amber to red with no intermediate state — an event that is almost always the surfacing of something long known further down. Or the completion date is missed by a margin that no report ever forecast.
Two structural remedies
The first is a second channel. Somewhere in the governance arrangement there should be a reporting line that does not pass through delivery management: an independent review, an internal audit function, a technical assurance role, or an owner-side representative reporting directly to the sponsor. Its purpose is not to duplicate the primary report but to be answerable to a different party, so that the filter applied to it is a different filter.
The second is primary evidence. Alongside the narrative summary, the same audience should receive the underlying artifacts: the current schedule file rather than a milestone graphic, the open change log rather than a total, the risk register with owners and dates rather than a heat map, and the cost-to-complete build-up rather than a variance figure.
Primary evidence cannot be summarized optimistically because it has not been summarized. A reader who wants to test whether a program is credible can do so, and — more importantly — the authors know that the reader can, which changes what gets written.
Downward reporting matters too
The phrase is deliberate. Reporting that travels only upward also fails in the other direction: the site rarely learns what the board decided, or why.
That has a practical cost. Where a delivery team does not know which constraints are genuinely fixed and which were assumptions, it optimizes against the wrong ones — protecting a date the board would have traded, or accepting a scope reduction the business needed. Decisions taken at the top and not communicated downward are re-litigated informally at every level below, which is one of the more expensive forms of decision latency.
A reporting arrangement worth having therefore closes the loop. What went up was tested against evidence, and what came back down was the decision and its reasoning. Either half missing produces the same outcome: an organization that is confident about a project it cannot actually see.
