Statistics you can defend
Days-remaining projection
How long until enough traffic has arrived to settle the difference you are currently looking at.
The projection divides the sample still needed by the rate traffic has been arriving at. It is deliberately not a fixed estimate made at launch: it is sized against the difference currently being measured, so it moves as that difference moves. A smaller effect needs more traffic, and the number reflects that rather than hiding it.
The rate is measured over the window the experiment was actually live. Dividing by the time since it started would count the days since it ended as if no traffic arrived, which understates the rate and inflates the estimate a little further every time somebody opens the page.
It refuses to answer rather than guess. A finished experiment has no remaining time and says so. One with no traffic yet, or less than a day of history, reports that it cannot tell. And a target that current traffic will not reach inside a year is reported as unreachable rather than as a precise-looking four-digit day count, because a number like that reads as a plan.
Read it next to the detectable-effect floor. "Forty days" and "this difference is below what your traffic can resolve" are different situations with different answers, and only one of them is solved by waiting.