There is a type of project that fails silently. The report is green, the percentage of progress is comforting, the cost is in line — and then, three months from the end, a delay emerges that "no one had seen", a supply blocked, a cost that didn't add up. It's not bad luck: it's reporting built on data that wasn't there.
The red numbers, after all, are the easy problem: they are seen, discussed, decided. Missing data is the difficult problem: it doesn't appear in the report, doesn't trigger any decisions and turns green into an opinion. This article explains why data holes are more dangerous than data gaps, where they hide, and how honest reporting can reveal them instead of covering them up.
Because a missing data weighs more than a red number
A red number produces a reaction: convocation, analysis, action. Missing data produces silence: the report is compiled anyway, with the available data, and the result seems complete because no one reports that something is missing. The damage is not in the single hole: it is in the overall effect, a partial picture that is read, decided and archived as a whole picture.
There is a second, more subtle effect: the holes tend to concentrate precisely where the problems are arising. Undetected progress is often that of businesses in difficulty; outdated procurement hides late orders; unrecorded costs arrive unexpectedly at accounting closing. The silence of data is not neutral: it is selective, and it selects against you.
The three typical holes
Procurement not updated.Orders issued but not recorded as commitments, deliveries expected but not tracked, contract changes stuck in someone's inbox. The result: the final cost is underestimated and supplier dependencies do not appear in the plan.
Progress not detected."Sensation" percentages, activities marked as completed without verification, work packages that have been stopped for weeks because no one updates them. The result: the percentage of progress measures the latest reading, not reality.
Unrecorded costs.Invoices awaiting registration, hours of external resources not included, internal costs not allocated. The result: the final balance seems better than it is, until accounting arrives to bring everyone back to reality.
A concrete example
Example (dummy data). A project with a budget of €200,000, divided into 10 work packages of €20,000 each. The monthly report declares a progress of 70%: seven parcels closed, costs recorded at €90,000. All green.
In preparation for the review, the audit uncovers three things. First: three of the seven "closed" packages have testing still open — actual progress, 80% each. Second: two packages that have not been detected for three weeks are late. Third: a €15,000 supply order was not recorded as a commitment. The actual progress drops from 70% to 64% (four parcels at 100% plus three at 80%, out of ten), the cost to finish rises from €185,000 to over €210,000 and the completion date slips by three weeks. The report was green: data was missing, not problems. End of example.
The mechanism of the example is known from the literature: quantitative control techniques, such as earned value analysis, produce reliable predictions only if the input progress is measured and complete; on partial data, each prediction inherits the hole (a recent proposal combines earned value and Monte Carlo simulation to detect deviations).
How a report can declare missing data
The solution is not to have all the data - no project has ever had it - but to declare what is missing. In practice: a fixed section of the report, "Missing data and estimated impact", with a simple table. Here's an example (dummy data) of what this looks like:
| Missing data | Estimated impact | Who gets it back | I'll go in |
|---|---|---|---|
| Purchase order not registered | Cost to finish underestimated | Purchasing office | 3 days |
| Advancement of 2 parcels not detected | Uncertain progress percentage | Parcel manager | 1 week |
| Invoices for the month not entered | Partial final balance | Administration | 2 days |
The rule is simple: declared missing data is managed missing data; an unspoken missing data is an error that accumulates. And the report declaring the holes is more credible, not less: it shows that the measure is under control. Inproject progress report, this section naturally goes alongside the indicators: it is the part that says what the indicators don't say.
Checklist: the check before the review
Before each overhaul, a few minutes on these seven checks.
- Are all orders to suppliers recorded as commitments, including verbal orders?
- Does each work package have a progress measure for the last seven days?
- Do progress percentages have a measurable basis or are they guesstimates?
- Are all invoices received and hours worked included in the final balance?
- Have the "closed" activities passed a verification, such as testing or acceptance?
- Is the cost and time baseline updated to the latest approved change?
- Did anyone sign off on the accuracy of the data before the meeting?
If an answer is "no", you have found a missing piece of data: state it in the report, with the estimated impact, instead of leaving it out. The checklist does not serve to blame those who have not updated: it serves to make visible the boundary between measured and presumed.
What this means for your reporting
In my opinion, the most useful test for a project report is this: can those who read it distinguish true data from presumed data? If the answer is no, the report is a narrative, not a measurement. The distinction does not require sophisticated tools: it requires the discipline of declaring holes. The tools are used later, to make the distinction automatic: a reporting that connects each indicator to its source and signals when the source is old or incomplete, as explained by theproject analytics guide. You can also read on the same topichow to evaluate the progress of an innovation projectewhy complex projects fail.
In the revision, question 5 of the path onten questions of a review that decidesasks exactly this: what data is missing to decide.
The next step
Before your next review, do the seven checks on your most important project: a few minutes, a table to add to the report, one less risk that no one sees. If you want to see how structured reporting handles data provenance and updating, start with oneControlRoom free guided demo. The risk that no one sees can be seen very well, once you decide to look for it.