Better than 90% of users: 5 patterns of questions asked by DeepSeek experts!

Why are AI’s answers always unsatisfactory?

It's obviously the same problem, why do others' results are better than yours?

In fact, the problem is not with AI, but with the way you ask questions!

Today, I will share a "five-step method" framework for efficient questioning from DeepSeek released by Byte Notebook Planet, which allows you to easily master AI and solve problems like a master!

Clarify roles, tasks and format (RTF)

The first step is to tell DeepSeek who it is, what it is supposed to do, and in what form the results should be presented. This gives AI a clear "identity card" and "task list".

Bad example : "Help me write a product description."

Correct example : "Please play the role of a professional e-commerce copywriter (Role) and write an attractive product description for me (Task) in the form of a 500-word paragraph (Format)."

Analysis : This question clearly specifies the role that AI needs to play (Role e-commerce copywriting expert), specific tasks (Task writing product descriptions), and presentation format (Format500 field).

By clearly specifying these three elements, AI can quickly understand your needs and provide professional, targeted content rather than generalities.

Tip : Even though DeepSeek currently deals with text input and output, you can still request answers in a specific format, such as lists, tabular text, or structured content. Specifying a format can make the output clearer and more organized!

Tasks, Actions and Goals (TAGs)

Having a task alone is not enough, we also need to let DeepSeek know how to do it and what the ultimate goal is.

Draw a "roadmap" for AI.

Wrong example : "I want to make a short video, give me a promotion plan."

Correct example : "Please develop a marketing strategy for my short video (Task), including specific steps for topic selection, filming and promotion (Action), with the goal of attracting 100,000 views within 30 days (Goal)."

Analysis : This question clearly defines three key elements: specific tasks (Task: develop short video marketing strategy), required actions (Action: including topic selection, shooting and promotion steps), and quantifiable goals (Goal: 100,000 views within 30 days). Such questions allow AI to provide targeted and practical suggestions instead of empty marketing theories. Specific numerical goals also allow AI to adjust the radicalness of the recommended strategy according to actual conditions.

Tip : The more quantitative your goal is, the more precise DeepSeek's answer will be. For example, "attract more traffic" is not as clear as "increase 10,000 followers in 30 days."

Context, Action, Outcome, and Example (CARE)

DeepSeek's logical reasoning ability is very powerful, but it needs enough background information to play its best role. This step is like providing AI with "puzzle pieces" to let it piece together the picture you want.

Wrong example : "Write an article about environmental protection."

Correct example : "I need to write an article about environmental protection for Toutiao (Context). The goal is to call on readers to participate in garbage sorting (Action). I hope that the article is interesting and persuasive, and ultimately attracts at least 50% of readers to like it (Result). Please refer to the style of XXX official account (Example)."

Analysis : This question provides complete context information (Context is writing environmental protection articles for the official account), clarifies the action goal (Action calls on readers to participate in garbage sorting), sets specific result expectations (Result is interesting, persuasive, 50% like rate), and provides a reference example (Example is a specific official account style). Such comprehensive background information allows AI to accurately grasp the writing direction and style, and create content that not only meets your expectations but also resonates with readers.

Tip : DeepSeek is good at reasoning based on context. The more detailed the data you provide, the more tailored its answer will be to your needs.

Before, After and Bridge (BAB)

Sometimes, our questions are very complex, and asking them directly may make DeepSeek unable to solve them. In this case, you may want to break the question into "before" and "after" and let DeepSeek help you find the "bridge" to the solution.

Wrong example : "How to do SEO optimization?"

Correct example : "My website currently ranks 30th in the search engine (Before), and I hope to enter the top 10 within 90 days (After). Please help me develop an SEO optimization strategy (Bridge)."

Analysis : This question clearly describes the current situation (Before ranking 30th) and the expected result (After entering the top 10 within 90 days), and clarifies the solution that AI needs to provide (Bridge SEO optimization strategy). This breakdown allows AI to provide a practical step-by-step improvement plan based on the current situation and the gap between the goals, rather than general theoretical suggestions. The time constraint (90 days) also allows AI to provide suggestions that fit this time frame.

Tips : If the problem is too complicated, you can ask the questions step by step and let DeepSeek help you analyze the current situation first, and then develop a plan.

Let’s move on to more advanced operations: Roles, Inputs, Steps, and Expectations (RISE)

For more complex tasks, we need to further refine DeepSeek’s role, input information, execution steps, and expected results, giving AI a complete “operation manual”.

Bad example : "Help me come up with a content plan about environmental protection."

Correct example : "Please play the role of a content marketing expert (Role role), based on the keyword 'environmental living' (Input input) I provided, develop a 30-day content plan for me, including topic selection, writing and publishing steps (Steps steps), hoping to attract at least 5,000 readers (Expectation expectation)."

Analysis : This question assigns AI a professional role (Role content marketing expert), provides clear input materials (Input environmental protection life keywords), details the required execution steps (Steps topic selection, writing, publishing), and sets measurable expected results (Expectation attracting 5,000 readers). This comprehensive guidance allows AI to exert its professional capabilities and provide systematic and actionable solutions based on specific domain knowledge, rather than simply listing content ideas.

Tips : DeepSeek supports iterative optimization. If the first answer is not ideal, you can add more information to let it adjust the result.