That’s a difficult question to answer! Experiment design is crucial for good research but each one tends to be different!
The unifying theme is that you must know what changes you can measure during the course of an experiment. So if you have a way of measuring compound A and no longer way to easily measure compound B, and you want to run an experiment when A turns into B, you may want to look at the disappearance of A as an indicator of the reaction proceeding properly. Of course you then have to make sure the product you gave is indeed B!
Once you know what changes you can measure you want to work out what this change would mean and how you could use it to test a hypothesis. An example would be if you wanted to find some catalyst that decomposes hydrogen peroxide. If you know that when hydrogen peroxide decomposes it creates gas (oxygen), you could simply take several tubes of the peroxide and try dropping different compounds into the separate tubes and see which makes bubbles appear. If you can measure how many bubbles and how much gas comes off you can also tell how active this catalyst is.
The best kind of experiments are ones which will give you a direct answer to a question regardless of how they turn out. A win-win scenario!