Can you predict a missed deadline before it happens?

Leading versus lagging

Almost every project measure in common use is lagging. Burndown describes work completed. Velocity describes past sprints. A missed milestone is the most lagging indicator there is — it reports something that has already finished happening.

A leading indicator moves before the thing it predicts. In our research, an engineering leader named the gap exactly:

"Have seen things slip. It would get better if we had some kind of leading indicator."

Team confidence qualifies, for a structural reason rather than a mystical one. The people doing the work encounter the evidence first — the API that doesn't behave as documented, the requirement that turned out to be three requirements, the test suite that's slower every week. Their belief about the deadline updates as they meet that evidence. The official date doesn't update until someone decides to escalate, and escalation is expensive, so it lags by weeks.

The distance between those two moments — belief changing, and the date changing — is the space a confidence signal operates in.

Why the gap exists at all

It would be reasonable to ask why, if the team knows, the date doesn't move sooner. Three reasons, all documented in our interviews.

Certainty arrives late. Early doubt is unprovable. "I think this is going to be harder than we thought" invites the question how much harder?, and until you can answer, raising it costs credibility.

Escalating is expensive. A project manager described the mechanics:

"Switching a ticket from green to yellow is such a big step change that it sets off all the alerts. There is a cost to calling out problems."

Nobody carries a baseline. A team whose confidence erodes over six weeks doesn't experience erosion — it experiences six ordinary weeks, each marginally worse than the one before, none different enough from its predecessor to remark on. Slow change is invisible from inside without a record to compare against.

The consequence is the pattern a head of engineering described:

"I trusted what I was told by the dev manager. Two weeks before the summit, I was told they weren't going to deliver everything that was promised."

The team's belief had almost certainly moved weeks earlier. Only the announcement was two weeks out.

How much warning does this actually give?

The honest answer has to distinguish two things, because conflating them is how this gets oversold.

What it can't do: tell you a project will slip by eleven days, or give you a probability of on-time delivery. That would require base rates from comparable completed projects — reference-class forecasting — and neither Genchi nor the team holds that data. Anyone offering you a percentage likelihood of hitting a date is modelling something they can't observe.

What it can do: tell you that the people doing the work have become less confident than they were, and how quickly. That's a statement about belief, and belief is observable.

In practice this converts "we found out two weeks before" into "we noticed the drift while it was drifting." How much earlier depends on cadence and on how gradually the situation deteriorated — a weekly check-in can't surface anything faster than weekly, and a problem that appears on Tuesday and is understood by Thursday won't produce much of a trend.

The useful framing isn't days of warning. It's the difference between finding out while options are still cheap — adjust scope, add help, move the date deliberately — and finding out when the only remaining options are weekend crunch, cutting scope at the wire, or shipping something nobody's proud of.

What the shapes tell you

A leading indicator is only useful if you can read it. Briefly, and covered properly in what to do when the trend goes down:

• Gradual decline over several check-ins is the classic pre-slip pattern — something going wrong slowly that nobody has yet named.
• A sharp drop means a discrete event happened, and somebody knows what it was.
• One person well below the rest is the earliest signal of all, and the most frequently ignored.
• Movement in general is normal. Problems get found and solved; a signal that never moves is a signal nobody is taking seriously.

Where it fails

Three honest limits.

Collective error. If everyone on the team is confident about something that fails anyway, this measures their confidence faithfully. Shared optimism is real, and a measure of belief is not a measure of reality. This is the significant limitation, and no amount of aggregation removes it.

Distant deadlines. One interviewee identified this precisely: "If the goals are too far away, it might be green until it turns red." Confidence about something eighteen months out is weakly informative — nobody has met the evidence yet. The signal sharpens as a deadline approaches, which is a real constraint on long-horizon work.

Unclear goals. If the team doesn't share an understanding of what success is, the confidence rating measures several different questions at once and the aggregate means little. One leader called this "a pre-round: is the goal clear or not?" It's a genuine precondition.

How Genchi does this

Genchi asks every team member the same question at your chosen cadence: how confident are you that we'll achieve our goal? One to five, about two seconds, through Slack, email or the app.

Responses are aggregated before anyone sees them, so people can answer honestly. Each initiative shows the current score, the distribution, and the trend across recent check-ins — so a decline is visible as a direction rather than as a number you'd have to remember last week's version of.