I don't really want to give examples as I am looking for a general rule that I can apply to any behavior or tendency to determine whether it is normal or a disorder.
You can think of "abnormal" in two basic ways:
Statistical abnormality - The extent to which an individual is high or low on some trait compared to the average of the population. This use of the term is value-neutral, meaning that there is nothing inherently good or bad about being close to the average (i.e., "normal") or far from the average (i.e., "abnormal"). For example, an individual with an extremely high IQ is "abnormal" in the same way that someone with an extremely low IQ is "abnormal". In short, statistical normality is just a way to mathematically describe how extreme a value is compared to some reference. The key to this type of "normal" is that the reference point (e.g., mean) is empirically derived. It is simply a numerical description.
Social or clinical abnormality - The extent to which an individual expresses a value on some trait compared to some reference defined by people. This reference point (or set of points) might be defined by society (i.e., social norms) or by experts like clinical psychologists (i.e., diagnostic criteria). The key to these types of "normal" is that the reference point is not value-neutral. In fact, the reference points with social and clinical normativity are usually what is considered "good" by those setting the criteria.
People (professionals included) often fail to distinguish between statistical normality and social/clinical norms, but its important to keep these straight when thinking about mental or behavioral disorders. Saying that an individual is "abnormal" in a statistical sense means something very different than saying that an individual is "abnormal" in a clinical or social sense.
To answer your question:
Your question is about clinical normativity. There is one especially useful criterion that clinical psychologists use to determine whether a patient is presenting as abnormal (or "disordered"): The extent to which a patient's symptoms interfere with everyday functioning and basic well-being. An individual might meet several specific criteria for a disorder (e.g., ADHD, phobias), but if he or she has found a way to maintain stable relationships, keep a job, and is otherwise happy with their life, a clinician might decide against giving a formal diagnosis. In short, the clinician basically asks the patient to identify the things making them miserable and tries to determine if any of the symptoms might be responsible. Things that will tend to push a clinician toward making a diagnosis of a disorder are things like not paying bills, poor social relationships, not leaving the house, drinking too much alcohol, inability to stay employed, feeling anxious or depressed, etc.
Of course, the "everyday-functioning/well-being" heuristic isn't perfect, but it is very useful for helping to organize one's thinking about what makes a behavior or disposition "disordered". The main drawback to this heuristic is that there are many things that interfere with everyday functioning that aren't considered disordered, like grief. Grieving over the death of a loved one can be incredibly disruptive to a person's life, but it isn't actually clear that a magic pill that could prevent the grieving process would be a good thing. Is it good for someone to lose their spouse and carry on like nothing happened the next day? Reasonable people can disagree on the answer to this question, but (for now) most clinical psychologists don't consider grieving to be "disordered" unless the person has been grieving for a very long time (e.g., years). So, the "everyday-functioning/well-being" heuristic is generally useful when trying to figure out if something is disordered, but keep in mind that there are many situations in which its application is of limited value.