All project work is driven by underlying uncertainty from people, processes, technology, and tools. Let me restate for clarity Certainty is an Illusion, if you're looking for certainty, you're not going to find it in the project domain. You're not going to find it anywhere in the real world.
So the big question is not, how can we manage in the presence of uncertainty? There is only one way. OK, there are two ways.
The first way is to ignore the uncertainties and pretend that what happened in the past is going to happen in the future. And what might happen in the future can be controlled when it happens, or we can just ignore it and keep moving open. We can also pretend that small samples of what happened in the past are sufficient to forecast the Happy Path to what is going to happen in the future future. Both of these are a naive from a management perspective, and they are mathematically wrong.
Now a second way. Since all project work is driven by underlying uncertainty, the result is a probabilistic outcome of anything we do. Like the picture below. Each activity is governed by its probability distribution. When these activities are interconnected, or even when they are serial, the result is a new probability distribution for the STOP date and cost. As an aside, these distributions are not multiplied. There is no multiply operator between probability distributions. The numbers that are produced by the distributions, the possible values that can result, are represented by Integral or Differential Equations. Same for simple linear series of work so popular in some agile communities trying to apply Little's Law. Which by the way requires each random variable to be independent and identically distributed (I.I.D.). Which in any non-trivial project is simply not the case.
The actual schedule for the notion example above, with some defined probability distributions and the applicable of a Monte Carlo Simulator produces this probabilistic outcome for a deliverable. The tool that produces this outcome is RiskyProject from Intaver. There are several products like this, but this is my favorite. It starts with the Risk Register for both reducible and irreducible risks, all the probability setting, including residual probability after mitigation, connects these risks to the work in the Integrated Master Schedule, and produces charts like the one below, showing the probability distribution function and the cummulative distirbution function for a specific deliverable. For example this shows a 79% chance of completing on or before March 13, 2018 for the item in the schedule.
So since we shouldn't or actually can't ignore the underlying probabilistic behaviour of our projects elements. what can we do? Well we can manage in the presence of these uncertainties.
The basis of all decision making in the presence of uncertainty is estimating. If you have no uncertainty, making decisions is easy, in fact trivial. If you have a project in which decisions can be made without the help of estimates, it is likely that project is itself easy and possibly even trivial. Those types of projects have little or no uncertainty.
Assuming your project is not trivial, here's one way to manage in the presence of uncertainty.