Experiments in general are very clearly defined: They are a process that is designed specifically to answer a question. The experiment design has a very great influence on how much of an answer you get, and on how general the answer can be - if you are doing a chemical experiment with, say, helium, you cannot tell things about what would happen with oxygen, or hydrogen in place of the helium.
Experiments, to be called such, also have to be documented clearly, accurately and objectively enough to allow the experiment to be repeated - and the outcome, if things are done correctly in each instance of the experiment, should be the same or at least very similar. This means anybody can, theoretically, replicate a given published experiment to test the reliability of the experiment itself and the answers it gave.
Which means that a proper archaeological experiment needs to be objective, it has to be repeatable, has to be designed to answer a specific question, and the design of the actual process has to be suitable for this answer - you cannot say something about oak working if you used beech, and you cannot say something about the suitability of old tools if you use modern ones. It also has to be documented accordingly. Everything else is not archaeological experiments - it can be tests of tools and techniques, or archaeotechnic, or playing around, or making replicas, but it's not an experiment.
Now, very obviously, as soon as you go into archaeology, things get a little messier than in a nice, clean chemistry lab or physics facility, where you have controlled environments, controlled substances, and measurable outcomes.
The materials for archaeological experiments are very often natural stuff - and there can be a huge difference between wood and wood, even if it's the same tree species. Plus there's the human factor - craft skills are an important factor, and they are hard to quantify. Many things are also subjective, such as the preference of one style of working over another, or one tool variant over another. Which means that "objective" will, in various instances, turn into "as objective as possible" and "repeatable" into "as repeatable as possible" - since you cannot use the same tree twice. (In some cases, modern stand-in materials can be used to get better repeatability. I've learned about experiments about flint-knapping, for instance, that use specifically made ceramic cores for knapping: similar characteristics for the breaking, and uniform without flaws.)
One of the key things to still get workable results is... documentation. There's no such thing as too much documentation! This is especially important when human factors come into play. It's also important to keep in mind that these human influences may be very significant, and to document (and describe, and publish) accordingly. Even if it's not a proper, full-fledged archaeological experiment, but someone or a group trying out techniques, good documentation of the tests can make a world of a difference when you are trying to make sense of things - or just trying to get an inkling about a single person's skill development.
A very good example of documentation for this can be found on Agata Ulanowska's site. She's lecturing in Warsaw and has been giving courses about old textile techniques for several years, teaching students how to do things like tablet-weaving. For those courses, she's giving out documentation sheets, making it possible to track the progress of each student as well as getting an overview about times for designing, weaving, and finishing projects. The sheets are downloadable from her page (scroll down to find them) - maybe they will be helpful for you as well!