A date is a range, not a promise.
Ask when something ships and you get one date. That date is a guess wearing a suit. Here's what we show instead.
Every project has a moment where somebody asks "so when does it ship?" and somebody else says a date. One date. It goes in a slide, the slide goes upstairs, and from then on that date is a promise nobody actually made.
The person who said it knew it was a guess. They had a range in their head: probably the end of the month, maybe the week after if the payments vendor drags its feet. The range is the honest answer. The single date is what's left after the honesty gets sanded off.
Estimates start as ranges
In Plannibble the person doing a bite estimates it, and the estimate is a range: two to four days, not three. That isn't vagueness. "Two to four" says more than "three" does, because it tells you how sure the person is. A bite that's "three to three" and a bite that's "one to six" are very different bites, even though both average three.
Nobody else's number replaces theirs. When Nib has history on similar bites, it shows a history-adjusted range beside the original with the reason ("bites over five days here have run long seven of nine times"), and the person chooses which to keep. It sits beside the human number. It never quietly overwrites it.
The plan is worked out three times
To turn a pile of ranges into a date, the forecast works through the plan three times: once with every bite at the quick end of its range, once at a likely point, and once at the slow end. That gives three readings of the same plan: a best case, a likely date and a high end.
The likely date leads, because it's the one to plan around. The other two stand beside it, always.
Every assumption is written down
A forecast is only as honest as the things it assumes. When a bite is out with another company and nobody has a date back, the plan still has to put something there. Plannibble assumes three working days, and it writes that on the drawing, in orange, right on the line that waits: no date back · 3d assumed.
That's the part a single date always hides. If you can see the assumption, you can argue with it, chase it or work around it. If it's buried in someone's spreadsheet, you find out about it at the demo.
A target is not a forecast
We keep two words apart on purpose. A target is the date a leader sets: what we're aiming for. A forecast is the date the work says: nobody owns it, everyone feeds it. We never call either one "the deadline", because that word lets a target pass itself off as a prediction.
When the two disagree, the plan says so in one sentence that leads with the date: Checkout beta is likely Mon, Nov 2, four working days past its target. Then it shows you which bite is holding it there.
What this buys you
Ranges feel less satisfying than a date for about a week. After that they're a relief. Nobody has to defend a number they never believed. The person who reports upward gets to say "likely the 2nd, no later than the 3rd, and here's what it rests on", which is a sentence people trust.
And when the date moves, it moves for a reason you can point to. That's the whole idea: not a date that never changes, but a date that's never a surprise.