The warning signs that reveal a project in trouble
Most project failures announce themselves in advance. Delivery signals include missing demos, an endless "90 percent done," and a growing pile of bugs. Relationship signals include staff turnover and vague answers to direct questions. The earlier you catch them, the cheaper they are to fix – don't wait for the final invoice.
Almost no project fails overnight. It slides – slowly, with warning signs that were there weeks or months before the disaster became official. The trick is reading them early, because a problem caught in week four costs a fraction of the same problem discovered at the planned launch. The signals fall into two groups: the ones visible in delivery and the ones visible in the relationship.
Signals in delivery
These are the signs that the build itself isn’t moving the way it should:
- Missing or watered-down demos. If nothing working can be shown regularly, there’s rarely much finished behind the words.
- “90 percent done” week after week. The last ten percent is often half the work. A number that doesn’t move isn’t progress.
- A growing pile of bugs. When the list of defects gets longer every week and fixes create new bugs, stability is slipping.
- Estimates that keep blowing up. If every task takes twice as long as promised, the underlying planning is wrong, not the individual tasks.
- No one can say what’s left. Without a concrete list of remaining work, there’s usually no real control.
- Scope grows quietly. Features get added without a decision, and suddenly the team is building something bigger than what you ordered.
Signals in the relationship
Just as revealing is how the collaboration feels. People and communication leak information long before the numbers do:
- Staff turnover on your team. When key people are swapped out mid-project, context is lost, and you pay for the ramp-up.
- Vague answers to direct questions. “It’ll work out” and “we’re roughly on track” aren’t status reports, they’re smoke screens.
- Silence and slow responses. A vendor that stops reaching out proactively often has something it would rather not share.
- Blame always lands on you. If every delay is your fault – your decisions, your responses – ownership of the delivery is missing.
- Meetings are about process, not product. When conversations circle around method and routines instead of what’s been built, the format is often hiding thin substance.
- Your gut says something’s wrong. Experienced clients feel it before they can put it into words. Don’t ignore it.
A traffic-light model for turning signals into action
A single bad week isn’t a crisis. It’s the pattern over time that counts. Use a simple traffic-light model to tell a rough patch from a trend:
| Level | What it means and what you do |
|---|---|
| Green | A single signal, an isolated event. Note it, keep watching, don't act in a panic. |
| Yellow | Several signals, or one that persists. Raise it explicitly, ask for a plan and an updated forecast. |
| Red | Several red flags at once, over time. A structured crisis meeting, a documented replan, a deadline – or escalation. |
A scenario
Say the demo gets canceled one week (green – it happens). The following week shows the same thing as last time, and the bug list has grown (yellow – now you raise it). The week after that, you get vague answers about why, a developer has quit, and no one can say what’s left (red). At that point, don’t hope it turns around – act. Ask for an honest replan, set a deadline for improvement, and document what you agree on.
What to do when it’s flashing red
Have a structured conversation, ask for an updated forecast and a realistic plan going forward, and write down what you agree on. Often the fastest route to the finish line is cutting scope – launching something smaller that works beats chasing everything that isn’t done. If improvement doesn’t come within the deadline you set, you need to weigh an external review, escalation, or ending the collaboration against the cost of continuing.
At Weapp we work with regular demos and open forecasting specifically so the signals show up early, while they’re still cheap to fix. Want a second opinion on a project that feels shaky? Take a look at our services or get in touch.
Frequently asked questions
What's the earliest sign a project is slipping?
Missing or watered-down demos. As long as working parts are shown regularly, there's something tangible to point to. When demos get canceled, replaced by status decks, or just show the same thing as last time, it's often the first sign the build isn't moving forward.
What does "90 percent done" week after week mean?
That the last ten percent – testing, edge cases, stabilization – was badly underestimated, and that it's actually often half the work. A project that's been at 90 percent for a month isn't nearly done. Ask what concretely remains, task by task, rather than for a percentage.
Is staff turnover at the vendor a warning sign?
Yes, especially if key people on your specific project are swapped out mid-course. Every swap means lost context and ramp-up time you often end up paying for. Occasional changes happen, but recurring turnover on your team is a signal of problems either in the project or at the vendor.
How do I tell a normal rough patch from a real problem?
By the pattern over time, not individual bad weeks. A project is allowed one heavy sprint. But when several signals show up at once and persist – missing demos plus a growing bug pile plus vague answers – that's not a rough patch, it's a trend. The traffic-light model helps you tell them apart.
What do I do when I see several red flags at once?
Have a structured conversation with the vendor, ask for an honest replan and an updated forecast, and document what you agree on. Set a deadline for improvement. If it doesn't come, you need to consider escalation, an external review, or, in the worst case, ending the collaboration.