Every experienced delivery leader has seen one. The dashboard is green, the steering committee is calm, the milestones appear broadly intact, and the project manager is reporting that the team is working through the usual challenges. Then, almost without warning, the project falls apart. A dependency that was “being managed” becomes critical. A data issue that was “under review” delays testing. A supplier risk that was “within tolerance” becomes a commercial dispute. The business suddenly admits it is not ready, users are not engaged, and the planned benefits now depend on a heroic recovery that nobody budgeted for.
This is the watermelon project: green on the outside, red in the middle.
It is one of the most dangerous patterns in enterprise delivery because it gives leaders the thing they most want when a major program is under pressure: reassurance. Unfortunately, reassurance is not the same as control. A green status can be useful, but only if it is earned through evidence. When it is produced by optimism, fear, poor governance, or weak telemetry, it becomes worse than no report at all because it delays intervention until the cost of recovery has climbed sharply.
At Arqvera, we see watermelon reporting as a system problem before it is an individual problem. It is easy to blame a project manager for softening bad news, but that misses the point. People report into the conditions around them. If the organisation punishes candour, rewards surface-level calm, buries decisions in committees, or treats escalation as failure, the dashboard will eventually learn to lie politely.
The purpose of a project dashboard is to help leaders see enough truth to make timely decisions. It is not supposed to be a comfort blanket, although some governance forums have clearly taken a wrong turn there.
The problem is that many project status reports measure the wrong things, or measure the right things too late. They track plan adherence, spend, activity, milestone completion, and risk counts. Those indicators matter, but they do not always reveal whether the project is still viable, whether the business is ready, whether decisions are being made quickly enough, whether users are adopting the change, or whether the benefit case remains intact.
A project can be green on schedule because the team has deferred difficult scope. It can be green on budget because costs have not yet surfaced. It can be green on risk because nobody has updated the risk register honestly. It can be green on delivery because the system build is progressing while the operating model, data, process, and adoption work are all deteriorating underneath.
This is particularly dangerous in complex transformation. PMI’s 2026 Pulse of the Profession research found that 97% of project professionals managed at least one complex project in the previous year, with more than half of all projects classified as complex. PMI also found that around one-third of complex projects fail, nearly twice the failure rate for projects overall, and that project professionals who manage complexity effectively increase the likelihood of success by five times.
That matters because complexity changes the value of reporting. In a simple project, progress against tasks may tell you a great deal. In a complex transformation, the bigger risks sit in the connections between workstreams, decisions, incentives, data, suppliers, operating models, and people. If your dashboard cannot see those connections, it will report calm while the system is becoming unstable.
The uncomfortable truth is that people often green-shift reports for understandable reasons. They may not be trying to deceive anyone. They may be trying to survive the governance system they have been placed inside.
The behavioural research behind this is well established. The “Mum Effect,” first described by Rosen and Tesser in the 1970s, describes people’s reluctance to transmit bad news. Later research by Bond and Anderson found that this reluctance is often driven less by kindness toward the recipient and more by the sender’s desire to avoid negative association with the message. In other words, people avoid being the carrier of bad news because organisations have a habit of shooting the messenger, even when they claim to value transparency.
In project environments, this becomes very practical. A delivery lead does not want to be seen as negative. A supplier does not want to trigger commercial scrutiny. A workstream owner does not want to admit that their team is behind. A sponsor does not want to take a red status to the board unless there is a recovery plan. The PMO does not want to escalate every weak signal and be accused of crying wolf. The result is not necessarily one big lie. It is a hundred small acts of softening.
“Blocked” becomes “in progress.”
“No owner” becomes “awaiting alignment.”
“Users are not engaging” becomes “stakeholder management ongoing.”
“The data is unusable” becomes “data quality remediation underway.”
None of those phrases is automatically dishonest. The problem is that they often hide the decision the organisation needs to make. A good status report should not just describe activity. It should expose whether the work is still on a credible path to the outcome.
One of the reasons watermelon projects survive for so long is that organisations over-trust the dashboard. They assume that because the governance pack exists, reality has been captured. It has not. A dashboard is only as reliable as the signals feeding it and the culture interpreting it.
This is where systems telemetry becomes important. Leaders need more than reported status. They need operational signals that show how the project is actually behaving. Those signals may include decision latency, issue ageing, dependency slippage, change request volume, unresolved design questions, defect trends, data remediation burn-down, test readiness, user engagement, sponsor availability, supplier turnover, meeting effectiveness, benefit confidence, and the number of times the same risk has been reworded rather than resolved.
Some of the most useful indicators are behavioural rather than technical. Are meetings becoming longer but less decisive? Are senior people attending but not making calls? Are risks being carried from one governance cycle to the next without intervention? Are teams escalating informally because formal routes are too slow? Are workstream leads using increasingly careful language? Are the same three people holding the whole program together through personal effort?
That last one matters. Heroics are often the first sign that the system is failing. A project that depends on a small number of exhausted individuals to compensate for weak governance, unclear ownership, or supplier underperformance is not healthy. It is borrowing confidence from people who cannot sustain the loan.
There are several warning signs that a green project may already be in trouble.
The first is unresolved decision latency. If key decisions are repeatedly deferred, the project may still report progress for a while, but the delay will eventually show up as rework, cost, scope pressure, or compromised adoption. Decision latency is one of the clearest early indicators of governance weakness because it reveals whether authority is actually where the governance model says it is.
The second is dependency drift. Complex projects rarely fail because one task is late. They fail because dependencies between teams, suppliers, systems, data, processes, and business functions are not actively managed. If dependencies are noted but not owned, the project is building hidden risk.
The third is repeated re-baselining without a change in behaviour. Re-baselining can be legitimate when assumptions change, but if the project resets the plan without changing the operating model that caused the drift, the new baseline is just a more recent version of the same optimism.
The fourth is persistent ambiguity in ownership. If nobody can clearly say who owns data quality, process design, adoption, business readiness, benefit realisation, or supplier decisions, then the project may be structurally under-governed. Ambiguity is comfortable early and expensive later.
The fifth is a gap between delivery progress and business readiness. This is common in technology programs. The platform build advances, environments are prepared, configuration is demonstrated, and technical milestones are met, but business users remain under-engaged, process impacts are unclear, training is not connected to real role change, and operational leaders are not prepared to own the new ways of working. Technology activity is then mistaken for transformation progress.
The sixth is declining trust in the reporting itself. When leaders start holding side conversations to find out what is “really going on,” the formal reporting system has already lost authority. That is a serious signal. Informal intelligence is useful, but if it becomes the only route to truth, governance is no longer working.
The Association for Project Management defines governance as the framework of authority and accountability that defines and controls the outputs, outcomes, and benefits from projects, programs, and portfolios. It is also the mechanism through which the investing organisation exerts financial and technical control over the work and the realisation of value.
That definition should be read operationally. Governance is not the monthly meeting. It is the system that makes sure the right people see the right evidence early enough to make the right decisions. If governance slows down truth, it is not governance. It is theatre.
Good governance makes bad news travel faster and with less drama. It creates agreed escalation thresholds, clear decision rights, honest assurance points, and a culture where red is treated as a call to action rather than a career-limiting event. It also protects the delivery team from having to manage executive emotions before they can manage the project.
This is where sponsorship matters. PMI has long identified actively engaged executive sponsors as the top driver of project and program success, while also noting that fewer than two-thirds of projects and programs have assigned executive sponsors. A sponsor who only wants positive news will eventually receive it, regardless of reality. A sponsor who asks better questions, removes obstacles, and rewards candour gives the project a much better chance of recovering early.
Leaders do not need to become project managers, but they do need to ask questions that reveal system health rather than reporting polish.
A useful first question is: what would make this status turn amber or red? If the answer is vague, the RAG rating is probably subjective. Good reporting should have clear thresholds and leading indicators. If green is based mainly on confidence, personality, or recent momentum, it deserves challenge.
The second question is: where are we relying on individual heroics rather than process? This exposes whether the project’s apparent health depends on a few people absorbing unsustainable friction.
The third question is: which decisions are ageing? Aged decisions are often more dangerous than aged risks because they show where the organisation is failing to exercise authority.
The fourth question is: what does the business need to do differently after go-live, and are we on track for that behaviour change? This moves the conversation from technical delivery to adoption and value.
The fifth question is: what evidence would contradict the current green status? Mature governance actively looks for disconfirming evidence. Weak governance looks for reassurance.
The sixth question is: what are people saying outside the meeting that they are not saying inside it? This is not an invitation to gossip. It is a recognition that psychological safety and formal reporting quality are connected.
Independent assurance can feel uncomfortable because it interrupts the neatness of the program narrative. That is partly the point. A good assurance review is not there to admire the dashboard. It is there to test whether the dashboard reflects operational reality.
Arqvera’s Trust Arq provides independent delivery assurance across governance, readiness, partner performance, risk, and value confidence. It helps leaders understand whether a project is genuinely healthy, whether the right decisions are being made, whether the business is ready, and whether benefits remain achievable. The purpose is not to create more bureaucracy. The purpose is to create earlier truth.
For organisations that want an initial read on delivery confidence, Arqvera’s Project Health Self-Assessment is a practical starting point. It helps leadership teams identify where confidence is evidence-based and where the program may be relying on optimism, informal workarounds, or incomplete reporting.
This distinction matters. Many organisations discover the truth only after the cost of recovery has become severe. Earlier assurance gives leaders more options. They can reset governance, strengthen sponsorship, adjust scope, improve supplier management, re-sequence activity, add capability, pause a milestone, or reframe the business case before the program becomes too politically committed to admit what everyone already suspects.
Independent assurance is valuable, but organisations should also improve their own delivery telemetry. The aim is not to build a surveillance machine that buries teams in metrics. The aim is to identify a small number of leading indicators that reveal health earlier than traditional status reporting.
For example, a complex transformation might track decision latency, dependency age, unresolved design questions, business readiness completion, test defect severity, data quality confidence, adoption risk, sponsor availability, change request trend, benefits confidence, and supplier resourcing stability. These indicators should not replace judgment, but they make judgment better.
The important point is to combine human intelligence with operational evidence. A delivery lead’s instinct that something is wrong is useful. A rising defect trend is useful. A sponsor’s repeated absence is useful. A growing backlog of unresolved process decisions is useful. The signal becomes powerful when these are viewed together.
This is also where AI may become relevant, although not in the magical way some vendors imply. AI can help identify patterns across large volumes of delivery data, meeting notes, risk logs, issue trackers, and sentiment signals. It can highlight anomalies, repeated themes, and emerging clusters of concern. But AI will not fix a governance culture that punishes truth. The model may not be the problem. The organisation beneath it may be.
One of the most useful cultural shifts is to stop treating red status as failure. Red should mean that leadership attention is required. It should mean the project has surfaced a condition that needs a decision, intervention, or reset. In a healthy governance environment, red creates action. In an unhealthy one, red creates blame, so people avoid it until the evidence is overwhelming.
That is why the best organisations do not aim for permanently green portfolios. They aim for accurate portfolios. A well-run portfolio will show amber and red when reality requires it. It will resolve issues faster because it sees them earlier. It will have fewer nasty surprises because teams are not forced to maintain an appearance of calm. It will protect value because governance responds before delivery drift becomes business damage.
Accuracy is more useful than comfort. It is also much cheaper.
If a project is genuinely green, leaders should be able to explain why using evidence rather than mood. They should be able to show that decisions are being made, dependencies are being managed, the business is preparing to adopt the change, suppliers are performing, risks are ageing appropriately, benefits remain credible, and the governance system is helping rather than slowing the work down.
If those conditions are not true, the green status is not confidence. It is decoration.
Watermelon projects are not inevitable. They are produced by systems that prefer reassurance to candour, visibility to control, and activity to value. Fixing them requires better governance, better telemetry, stronger sponsorship, and a leadership culture that wants the truth early enough to use it.
Technology rarely fails in isolation. Projects rarely fail in isolation either. They fail inside the reporting, governance, incentive, and decision systems that surround them.
The question for leaders is simple: is your dashboard telling you the truth, or just telling you what your organisation has trained it to say?
Is an AI and technology transformation consultancy and advisory.
We help organisations shape business cases, projects, deliver excellence, and realise change and outcomes that stick. We support organisations before, during, and after projects with an end-to-end service where our domain specialization comes to life.
Before (Inception): We work with you to clearly define the idea, vision, strategy, and business case for change, as well as help select the right partners, and establish governance
During (Execution): We help deliver project and change objectives while keeping implementation under control through structured governance and assurance to realise intended outcomes.
After (Value Realisation): We ensure outcomes deliver measurable value and embed continuous improvement from successes and learnings.
Arqvera is led by industry veterans in the UK and USA with 100+ years of technology delivery intelligence across global consulting, digital transformation, and mission-critical projects and programmes.