Define beforehand what result you want to obtain, be it on paper or in a document, planning is always important, deconstruct from the broad concept to the detail.

Identify the context that explains the origin of where you will get the result, think about what is the most detailed and concise explanation for a friend in a bar or a child at school, explaining is an art.

Research the format of the result you will need to implement your prompt, help yourself with the expertise of a developer, or consult the best way to get what you really want.

Be concise and specific so that the model can understand what you are asking for.

Provide a broad context so that the model can generate a relevant and coherent response, yet always be precise to avoid double meaning.

Be consistent by specifying the format you want the result to be, besides playing with the different commands of the model, input and output, it is necessary to be punctual in what you want to get inside the prompt.

Learn or know beforehand the scope of the model and the provider you will use, one is not 100% equal to the other and can affect the construction of your prompt and the results you get. Be ahead of the curve!