I'd advise against adding some type of margin to your targets. It seems a lot of folks like to do this, but it is unreliable and just adds unnecessary costs to your budget.
First, there is no such thing as an accurate or correct target. Your estimate should ALWAYS be a range. For example, for task A, your workers estimate that this work will take between 12 and 24 hours to complete. Perhaps a most likely might be 16 hours. That is the probabilistic density function of that task in hours. If you did this work 100 times, this is the range of results you would likely experience. So as you can see, picking a single number in that range has a very low likelihood of occurring on any given attempt.
Adding a margin is about risk. Using the example above, you would simply push your targets on the outside of that range in the event the work is deemed high risk. So, if high risk you might target 17, 18, even 20 hours to do the work. By doing so, you are putting more of that density function behind you and, thus, giving you reserves to cope with uncertainty. Conversely, if the environment is low risk, you might target 16, 15, or 14 hours because of your confidence.
So instead of using some frozen margin to add to your work that has no basis in reality, estimate your range and understand the probability of those results, conduct a risk assessment on the environment, and then target accordingly.