General Tips for Designing Prompts

Here are some tips to keep in mind when designing your prompts:

Start Simple

When you start designing prompts, keep in mind that this is an iterative process that requires a lot of experimentation to achieve optimal results. Using a simple playground from OpenAI or Cohere is a good starting point.

You can start with simple prompts and add further elements and context as you aim for better results. Iterating your prompt over time is therefore crucial. As you read the guide, you will see many examples where specificity, simplicity, and conciseness often yield better results.

If you have a large task that includes many different subtasks, try breaking it down into simpler subtasks and build up as you get better results. This avoids introducing too much complexity at the beginning of the prompt design process.

The Instruction

You can design effective prompts for various simple tasks by using commands to instruct the model what you want to achieve, for example “Write”, “Classify”, “Summarize”, “Translate”, “Sort”, etc.

Keep in mind that you also need to experiment a lot to see what works best. Try different instructions with different keywords, contexts, and data and see what works best for your specific use case and task. Usually, the more specific and relevant the context is to the task, the better. We will discuss the importance of sampling and adding more context in upcoming guides.

Others recommend placing instructions at the beginning of the prompt. Another recommendation is to use a clear separator such as “###” to separate the instruction and the context.

For example:

### Anweisung

Übersetze den folgenden Text ins Spanische:
Text: "Hallo!"

¡Hola!

Specificity

Be very specific about the instruction and the task you want the model to perform. The more descriptive and detailed the prompt is, the better the results. This is especially important if you are aiming for a desired outcome or style of generation. There are no specific tokens or keywords that guarantee better results. What matters more is a good format and a descriptive prompt. In fact, providing examples in the prompt is very effective for getting desired outputs in specific formats.

When designing prompts, you should also keep the prompt length in mind, as there are limits to how long it can be. Consider how specific and detailed you should be. Including too many unnecessary details is not necessarily a good approach. The details should be relevant and contribute to the task at hand. This is something you need to experiment with. We encourage lots of experimentation and iteration to optimize prompts for your applications.

As an example, let’s try a simple prompt to extract specific information from a text.

Extrahiere die Namen von Orten im folgenden Text.

Gewünschtes Format:
Ort: <komma_separierte_Liste_von_Firmennamen>

Input: "Although these developments are encouraging to researchers, much is still a mystery. “We often have a black box between the brain and the effect we see in the periphery,” says Henrique Veiga-Fernandes, a neuroimmunologist at the Champalimaud Centre for the Unknown in Lisbon. “If we want to use it in the therapeutic context, we actually need to understand the mechanism.""

Ort: Champalimaud Centre for the Unknown, Lissabon

The input text is adapted from Kwon (2023)¹.

Avoid Vagueness

Given the tips above about richness of detail and improving format, it is easy to fall into the trap of trying to be too clever with prompts and creating potentially vague descriptions. It is often better to be specific and direct. The analogy here is very similar to effective communication - the more direct, the more effectively the message comes across.

For example, you might be interested in learning the concept of prompt engineering. You could try something like:

Erläutere das Konzept des Prompt-Engineerings. Halte die Erklärung kurz, nur ein paar Sätze, und sei nicht zu beschreibend.

From the prompt above, it is not clear how many sentences to use and what style. You might still get some good answers with the prompts above, but a better prompt would be very specific, concise, and to the point. Something like:

Nutze 2-3 Sätze, um das Konzept des Prompt-Engineerings einem Oberschüler zu erklären.

Do or Don’t?

Another common tip when designing prompts is to state what to do instead of saying what not to do. This promotes more specificity and focuses on the details that lead to good responses from the model.

Here is an example of a movie-recommendation chatbot that does exactly what I do not want because of how I wrote the instruction - focused on what not to do.

Das Folgende ist ein Agent, der Kunden Filme empfiehlt. FRAGE NICHT NACH INTERESSEN. FRAGE NICHT NACH PERSÖNLICHEN INFORMATIONEN.

Kunde: Bitte empfehlen Sie einen Film basierend auf meinen Interessen.
Agent:

Natürlich kann ich einen Film basierend auf Ihren Interessen empfehlen. Welche Art von Film möchten Sie sehen? Bevorzugen Sie Action, Komödie, Romantik oder etwas anderes?

Here is a better prompt:

Das Folgende ist ein Agent, der Kunden Filme empfiehlt. Der Agent ist dafür verantwortlich, einen Film aus den globalen Top-Trendfilmen zu empfehlen. Er sollte davon absehen, die Nutzer nach ihren Vorlieben zu fragen und vermeiden, nach persönlichen Informationen zu fragen. Falls der Agent keinen Film zum Empfehlen hat, sollte er antworten "Sorry, ich finde heute keinen Film zum Empfehlen.".

Kunde: Bitte empfehlen Sie einen Film basierend auf meinen Interessen.
Agent:

Tut mir leid, ich habe keine Informationen für Ihre Interessen. Allerdings ist hier eine Liste der aktuellen globalen Top-Trendfilme: [Liste von Filmen]. Ich hoffe, Sie finden etwas, das Ihnen gefällt!

Some of the examples above were adapted from an article by OpenAI (2025)².

Better Prompts for better Results

Basic Prompt Variants

Text and Content Optimization