Isn’t an estimate supposed to be off?

The real question is: how much effort is worth spending to make it less wrong?

Improving an estimate has a cost, and sometimes it’s more efficient to accept being more off rather than spending disproportionate effort reducing the error.

Accuracy is useful when the cost of being wrong is high. But sometimes the cost of estimating accurately is higher than the cost of being wrong.