The completeness guarantee the whole planning apparatus rests on has no verification procedure attached to it, which is why the enterprises most confident in their plans are the ones least able to say what is missing.
Every planning method in general use rests on a single guarantee: that the decomposition of the work accounts for the whole of the work. Nothing missing, nothing double-counted, everything the project must produce present somewhere in the structure, including the effort of managing the project itself. The rule is stated in absolute terms wherever the discipline explains itself, and treated as the foundation beneath the schedule, the budget, the resource plan and the risk register.
There is no procedure attached to it. Not a weak one. None.
The evidence sits in the material the discipline uses to teach the rule. One worked decomposition presents seven top-level branches, one drawn with no children at all, two duplicating each other's subject, and none of them project management, in a document whose own list of common failures names forgetting project management as a classic omission. The same document states, of the same diagram, that each parent is fully decomposed by its children and the rule holds at every level. A second exemplar leaves a branch undecomposed, omits project management, omits production and launch, and is accompanied by the explicit finding that the structure satisfies the rule because every deliverable required to ship a marketable product is captured. A third is built from a scope statement containing a performance behaviour and a conformance obligation, neither of which appears in any branch, then runs a step formally labelled as the test against the rule and records that it passed. A fourth explains the rule by saying that a powerpack branch omitting the fuel system would violate it, and then presents a powerpack branch containing engine, transmission and cooling.
What survives across all four is the shape of the failure rather than any one error. The rule is asserted, the exemplar breaks it, and the compliance statement is written anyway, because writing it requires nothing to have been checked. For an enterprise, that is not a curiosity about teaching material. It is a description of what happens at its own baseline approval.
The Strategic Context
Work that was never decomposed is invisible to every control the enterprise owns. A cost variance requires a budget line to vary from. A schedule variance requires an activity to slip. A risk requires a register entry. Earned value requires a work package with a value to earn. Every instrument in the control room measures the plan against itself or against actuals derived from the plan, which means the control system's sensitivity to omitted work is exactly zero, on day one and on every day after.
Consider a hypothetical secondary steel fabricator taking a structural package on a process plant. The decomposition covers cutting, drilling, fitting, welding, blasting, painting and delivery. It is deliverable-oriented, properly coded, sensibly sized, and it passes review. What it omits is the connection design interface with the consulting engineer, the shop drawing approval cycle and its resubmissions, weld procedure and welder qualification records, compilation of material test certificates, fit-up rework at site, and coating touch-up after erection. Every one is work the fabricator will certainly do. None is a piece of steel, which is why none appears in a structure organised around pieces of steel.
The margin on that package was set from a total built out of the decomposition. The contingency was set as a percentage of that same total. The omitted work is therefore not covered by contingency in any meaningful sense, because contingency scales with what was found, not with what was missed. The enterprise carrying ten per cent against a base understated by more than that is not protected. It is precisely as exposed as it would be with no contingency at all, and considerably more confident.
What Leaders Misread as Rigour
The first misreading is that a mature controls environment implies a complete plan. Maturity increases internal consistency: identifiers reconcile, roll-ups foot, reports arrive on time, variances are explained. Every one of those improvements raises confidence without touching completeness. That is the mechanism behind the uncomfortable pattern in which the enterprises most certain of their plans are least able to say what is absent from them. Certainty is produced by the consistency checks; absence is not something those checks are shaped to find.
The second misreading is that experienced estimators would notice. Experience makes omissions more systematic rather than less. An experienced team starts from the structure that worked last time, and a reused structure carries forward every omission the original contained, now with the authority of precedent. Two decades of reuse produces a template whose gaps have never once been questioned, because each successive project inherited them as the definition of normal.
The third is that review catches it. A reviewer holding a well-formed decomposition sees well-formed branches, correct coding, sensible sizing and a clean roll-up, and reports accordingly. Nothing in that field of view indicates what is not there. Asking a competent person to look harder at the same document does not change what the document contains.
Reframing the Issue
The reframing that matters is arithmetic rather than cultural. The only completeness test the discipline ever specifies is to sum the children and confirm they equal the parent. That test cannot detect an omission, and the reason is structural: the parent's value is derived from the children. If a branch is missing, the roll-up is missing it too, and the two agree perfectly. Roll-up detects double counting and transcription error. It is arithmetically incapable of detecting absence, and it is offered as the completeness check.
Stated as a rule an executive can apply immediately: a completeness check that can be performed using only the plan can only detect inconsistency, never absence. Absence can only be found by comparing the plan against something outside it. Nothing in the standard apparatus requires that comparison, names who would perform it, says when it happens, or attaches a consequence to the finding.
This distinguishes cleanly by level. On a project, omission is a cost event. On a programme it becomes structural, because the same reused template is applied across every project and the same categories are missing from all of them at once. At portfolio level it becomes a systematic understatement of the enterprise's forward commitment, and the understatement is not random. It is concentrated in one direction, always understating, never overstating, because a template accumulates gaps and does not accumulate surplus.
Why the Only Test on Offer Cannot Work
The omissions are systematic, not random
The work that goes missing is the work that belongs to nobody in the room. Decompositions are built by people who own deliverables, and the recurring omissions are the things that sit between deliverables: project management effort itself, integration and qualification, certification and conformance evidence, commissioning, data migration, training and workforce ramp, interface work either side of a contract boundary, and decommissioning of whatever is being replaced. Anyone can predict this list in advance, which is the point. A failure this predictable is not a competence problem. It is an unassigned procedure.
A performance obligation is not a part
The second pattern is that decompositions handle components well and properties badly. A stated obligation that the product behave a certain way under load, or conform to a published standard, is not a thing that can be drawn as a box beneath a parent, so it is not drawn at all. The conformance work then arrives late, as a discovery, priced by whoever is holding it. A hypothetical national postal and parcel network modernising its sortation illustrates the same shape at scale: machinery, sites and software decompose cleanly, while the transitional double-running of old and new networks, the recutting of delivery round boundaries and the address data migration have no natural parent and appear in no branch.
Absence produces no variance, only surprise
Omitted work does not degrade performance indices, because it is in neither the numerator nor the denominator. It surfaces as a variation, usually at a point where the supplier has no competitive tension and the enterprise has no alternative. This article concerns work absent from every baseline; the separate failure in which the work is present in all four baselines and they still disagree about it belongs to [Related article: Four Baselines, One Project, No Referee]. Decision time is the largest single category of work that consumes schedule while carrying no cost line, and this article does not treat it, because it is the entire subject of [Related article: What Does It Cost You to Wait for a Decision?].
Decision Framework
The omission audit is not a review of the plan. It is five comparisons of the plan against documents that were not derived from it, run by someone who did not build it and does not report to whoever did. Each comparison produces a list of items with no home. It takes a small team two to three days on most projects.
| Reference document | What is compared | Finding |
|---|---|---|
| Commitment text: contract, business case, scope statement | Every obligation-bearing clause, read individually | Any clause mapping to no element |
| Predecessor project cost ledger | Every code that carried material spend last time | Any code with no counterpart this time |
| Day-one operating handover list | Training, spares, documentation, licences, data, staffing ramp | Any item no element produces |
| Exit list for what is being replaced | Decommissioning, migration, parallel running, disposal | Any item with no owning element |
| Interface list: every contract, entity and system boundary | Work required on each side of each boundary | Any boundary with an unowned side |
Three rules make it an audit rather than an opinion. The output is a list of unmapped items, never a judgement on plan quality. Every finding is dispositioned in writing: added to the plan, explicitly excluded with a named individual accepting the exclusion, or transferred to another party with the reference recorded. And it is a gate, not advice.
The thresholds are deliberately blunt. Any clause in the commitment text mapping to no element blocks baseline approval. Any predecessor cost code carrying material spend with no counterpart blocks baseline approval until dispositioned. A decomposition containing no project management element, or containing one expressed only as a flat percentage, fails outright, because a percentage is a price rather than a plan. Re-run the audit at any gate where the commitment text changes.
From Strategy to Execution
Immediate. Take the two largest plans currently in delivery and run the first reference only: read the contract clause by clause against the decomposition. This can be done in a day and it will produce findings. Publish the count of unmapped clauses, not a narrative. Separately, require every plan submitted for approval this quarter to name the individual who performed a completeness comparison and the outside document they compared against.
Medium term. Make the omission audit a condition of baseline approval, with the disposition record attached to the approval. Change who signs the completeness statement: it cannot be the person who prepared the plan, and it should not be a committee. Instrument the reuse problem by requiring every plan built from a prior structure to declare which structure it came from, so that when an omission is found it can be corrected in the template rather than in one project.
Long term. Treat the disposition record as a permanent asset. The value of the audit compounds only if findings accumulate into a standing list of categories the enterprise habitually misses, which is then read into the next decomposition as an input rather than discovered again as a variation. What the resulting duration does to project cost through overhead recovery is outside this article; it belongs to [Related article: The Cheapest Duration You Only Find When You Are Late].
Signals to Monitor
The plan contains no project management element, or carries one as a flat percentage. A branch survives to baseline with no children beneath it. Two sibling branches share a subject. The scope statement contains a performance or conformance obligation that no element produces. The estimate was copied from a prior structure and nobody can say which one. Early variations are dominated by work everyone agrees was always required. Contingency was fixed before the decomposition was finished. The completeness sign-off carries the signature of the person who prepared the plan. No plan in three years has been sent back for missing work.
Questions for the Leadership Team
- What procedure in this enterprise verifies that a plan contains all the work, who performs it, and when did it last return a finding?
- On our largest current project, how many clauses in the commitment text map to no element of the plan, and who has checked?
- Which cost codes carried material spend on the last project of this type, and which of them have no counterpart in the current plan?
- Where does the effort of managing this project appear in its own decomposition, and is it an estimate or a percentage?
- Which prior structure was this plan copied from, and what was found missing from that structure by the time its project closed?
- Of the variations approved in the last twelve months, how many were for work that everyone agreed had always been necessary?
Closing Perspective
The absence here is not a gap in an enterprise's process documentation. It is a gap in the discipline itself, inherited intact by everyone who practises it, and it will not be closed by better plans or better planners. It is closed by one deliberate act: appointing someone to compare the plan against the world, before the plan becomes a commitment, and giving that comparison the authority to stop an approval.
Until that appointment is made, every completeness statement an enterprise signs is an assertion about a fact nobody looked for. Leaders should be clear about what they are underwriting when they sign it. Not that the plan is complete, which no one has established, but that the plan is internally consistent, which is a much smaller claim and the only one the evidence in front of them actually supports.
About EraNorth Insights
EraNorth Insights publishes practical analysis on strategy, projects, operations, transformation and decision intelligence for professional and organisational use. About EraNorth.