Most poor AI results have a simple cause: the prompt was too vague. "Write me an email to the customer" delivers average results. If you specify the recipient, content, and length, you'll get a draft you can send almost as is.
In this article, you'll find 40 AI prompts for eight business areas, from marketing to IT. Each prompt is structured so that you only need to replace the placeholders in square brackets. The templates work in ChatGPT, Claude, Gemini, Microsoft Copilot, and similar assistants. At the end, we'll show you when a prompt is no longer enough and you're better off using a skill.
What makes a good AI prompt?
A good AI prompt tells the AI what task to perform and what material to work with. It also specifies who the result is for and what it should look like. The more concrete the goal, context, and format, the more useful the answer. Constraints like "don't invent any numbers" prevent typical errors.
All templates in this article follow this structure. Five building blocks appear again and again:
- Task: What exactly should be created, such as a draft, a table, or an evaluation.
- Material: Notes, texts, or numbers for the AI to work with.
- Target audience: Who the result is for, so that the tone and level of detail are just right.
- Format: Length, structure, and form, such as "maximum 150 words" or "as a table."
- Constraints: What the AI should not do, for example, inventing facts or mentioning prices.
How to further improve your prompts is explained in our article on Prompt Engineering. You can find special commands for ChatGPT in our list of ChatGPT commands.
How to use the templates
Copy a prompt, replace everything in brackets, and add your material. For important tasks, ask the AI to clarify if anything is missing. Always double-check the results before sending them out, especially when it comes to numbers and names.
A quick note on data privacy: Personal data belonging to customers, applicants, or employees should only be used in company accounts that have a data processing agreement in place, such as ChatGPT Business. In personal accounts, make sure to remove names and specific details beforehand. Find out more in AI and data protection.
AI prompts for marketing
In marketing, prompts are great for saving time on the first draft. You should always handle the final polish and approval yourself.
1. LinkedIn post from bullet points
Write a LinkedIn post for [Company] about [Topic]. Target audience: [Target Audience]. Use these bullet points: [Bullet Points]. The post starts with a concrete opening sentence, has a maximum of 1,200 characters, uses short paragraphs, and ends with a question for the readers. No hashtag lists, no emojis, and no fabricated numbers or client examples.
2. Content ideas from customer questions
Here are questions that customers frequently ask us: [List]. Derive 10 ideas for blog posts or social media posts from them. For each idea, provide a working title, the question the post answers, and the appropriate format: guide, checklist, comparison, or case study. Sort them by value for [Target Audience].
3. Newsletter draft
Create a newsletter draft for [Target Audience] on the topic of [Topic]. Material: [Notes or Links]. Provide three subject lines with a maximum of 45 characters, a preview header, and body copy of no more than 250 words. The text must end with exactly one call to action: [Goal, e.g., book an appointment].
4. Reviewing a landing page from a customer's perspective
Review the following copy of our landing page from the perspective of [Target Audience]: [Copy]. Answer: Can someone understand within five seconds what we offer and who it is for? What objections remain unaddressed? Which statements are too vague? Suggest concrete rephrasings for the three weakest parts.
5. Ad variations
Write 5 variations for a [Google or LinkedIn] ad for [Product]. Target audience: [Target Audience]. Main benefit: [Benefit]. Each variation uses a different angle: problem, result, number/stat, question, comparison. Adhere to these character limits: [Limits]. Do not use any claims or promises we cannot back up.
AI prompts for sales
In sales, prompts help with preparation, follow-ups, and practicing for tricky conversations.
6. Meeting preparation
I have a meeting with [Company]; the point of contact is [Role]. Here is what we know so far: [Notes]. Here is the text from their website: [Text]. Create a meeting preparation guide including three assumptions about their current challenges, five open-ended questions for the conversation, and two potential objections along with responses. Clearly label what is an assumption versus what comes directly from the source.
7. Post-meeting follow-up
Here are my notes from the conversation with [Company]: [Notes]. Write a follow-up email of no more than 150 words. It summarizes the three key points, outlines the agreed next steps along with dates, and ends with a concrete question. Tone: [Tone, e.g., friendly and direct].
8. Practicing objection handling
Roleplay as a skeptical head of procurement from the [Industry] industry whom we want to sell [Product] to. Raise the five toughest objections one by one and wait for my response each time. After each response, provide a brief evaluation: What was convincing, what wasn't, and how could I have formulated it better?
9. Scope of work for a proposal
Draft the scope of work for a proposal to [Company]. Client's initial situation: [Initial Situation]. Our service: [Services]. Structure into Initial Situation, Objective, Scope of Services, Process/Timeline, and Deliverables. Write objectively and without sales jargon. Leave out pricing and terms; I will add those myself.
10. Ideal customer profile
Here are our five best clients, including their industry, size, and the reason they bought: [List]. Based on this, derive a profile of our Ideal Customer (ICP): shared traits, typical purchase triggers, who the decision-makers are, and how we can identify matching companies. Conclude by outlining what specific details you are still missing to form a clearer picture.
AI prompts for customer service
In customer service, the right tone is everything. Good prompts always provide the AI with the facts of the case so it doesn't promise anything you can't deliver.
11. Responding to a complaint
A customer submitted this complaint: [Message]. Facts of the case: [Facts]. Write a response that specifically acknowledges the issue, explains what we are doing, and outlines a clear next step. Avoid empty phrases like "We apologize for any inconvenience" and make no commitments that are not backed by the facts.
12. FAQs from customer inquiries
Here are [Number] anonymized customer inquiries from recent weeks: [Inquiries]. Group them by topic and count how often each topic occurs. For each of the five most frequent topics, draft an FAQ question along with a concise, clear answer based on these product details: [Product Details].
13. Summarizing long email threads
Summarize this email thread: [Thread]. Structure into: What this is about, what has happened so far, what remains open, and who needs to do what next. Maximum ten lines. Quote commitments and deadlines verbatim.
14. Response templates for common inquiries
Create response templates for these five frequent inquiries: [Inquiries]. Each template should be at most 80 words, provided as both an email version and a shorter chat version. Tone: [Tone]. Mark placeholders that need to be customized on a case-by-case basis with [insert].
15. Checking the tone of a response
Review this customer response for tone and clarity: [Response]. Customer context: [Short description, e.g., upset long-time customer]. Point out sentences that sound dismissive or overly technical, and provide a revised version that preserves the original meaning and substance.
AI prompts for HR and recruiting
In HR, it is especially important not to use real names or sensitive data in private AI accounts. The following prompts don't require them.
16. Job posting
Write a job posting for [Position] at [Company] in [Location, remote, or hybrid]. Responsibilities: [Responsibilities]. Requirements, separated into must-haves and nice-to-haves: [Requirements]. What we offer: [Benefits]. Write concretely without buzzwords like "dynamic team," use gender-inclusive language, and keep it under 350 words.
17. Interview guide
Create an interview guide for [Position]. The three most important requirements are: [Requirements]. For each requirement, formulate two behavioral questions asking for a specific experience, such as "Tell me about a time when...". For each question, note what indicators characterize a strong answer.
18. 30-day onboarding plan
Create a 30-day onboarding plan for the [Position] role. Team: [Team]. Key tools: [Tools]. Initial tasks: [Tasks]. Structure it by Weeks 1 to 4, each with goals, scheduled meetings/milestones, and a responsible role. Keep names as placeholders.
19. Preparing for a feedback session
Help me prepare for a feedback meeting. My observations, without names: [Specific situations]. Goal of the meeting: [Goal]. Formulate an opening, the feedback for each observation as an "I" statement, and three open-ended questions. Focus the evaluation solely on behavior, not on the person.
20. Internal announcement
Draft an internal memo to all employees regarding [Change, e.g., new remote work policy]. Facts: [Facts]. The first two sentences state what is changing and as of when. These are followed by the reason and what each person needs to do. Maximum 200 words, clearly understandable without prior knowledge.
AI prompts for finance and controlling
When it comes to numbers, AI is a great tool for formulas, explanations, and reports. You should always double-check the calculations yourself.
21. Creating Excel formulas
I am working in [Excel or Google Sheets]. Column [A] contains [Content], and column [B] contains [Content]. I want to [Goal, e.g., calculate the sum of all outstanding invoices per customer]. Provide the formula, explain it step by step, and point out a common mistake to watch out for.
22. Explaining budget vs. actual variances
Here are the planned and actual figures for [Time Period]: [Table]. Identify the three largest variances, describe each in a single sentence, and list potential root causes I should investigate. Calculate the variances in both percentages and euros, and clearly flag any areas where data is missing.
23. Monthly report for management
Create a concise monthly report for executive management based on these figures: [Figures]. Structure: three key takeaways, a table comparing the primary KPIs against the previous month, and two areas requiring attention. Maximum one page, with no technical jargon left unexplained.
24. Payment reminder
Write a friendly yet firm payment reminder to [Client] for invoice [Number], dated [Date], in the amount of [Amount], which was due on [Date]. Include bank details as a placeholder and request payment by [Date]. Maximum 100 words.
25. Reviewing recurring costs
Here is a list of our recurring software and service expenses, including vendor, amount, and usage: [List]. Flag duplicate or rarely used line items, estimate the annual savings potential, and formulate one question I should ask the vendor for each of the three largest expenses.
AI prompts for projects and operations
This is where AI really shines at structuring: turning notes, conversations, and descriptions into plans, minutes, and guides.
26. Project plan
Create a project plan for [Project]. Objective: [Goal]. Deadline: [Date]. Team: [Roles]. Structure into phases with milestones, tasks, owners, and dependencies. Conclude by listing the three biggest risks and how we can identify them early on.
27. Meeting minutes
Here is the transcript or my notes from the meeting: [Text]. Create meeting minutes including participants, decisions, open items, and action items with assignees and deadlines. Include only what is stated in the text, and clearly mark any unclear responsibilities.
28. Documenting a process
This is how we currently handle [Process, e.g., order intake]: [Description]. Turn this into a step-by-step guide that a new colleague with no prior knowledge can understand. Number each step, specify the tool used for each, and highlight points where mistakes commonly occur.
29. Identifying automation potential
Here is a list of recurring tasks in our team along with the estimated weekly time spent on each: [List]. Evaluate each task based on how well it can be supported by AI or automation, and briefly explain why. Sort by potential time savings and outline a concrete first step for the top three candidates.
30. Comparing supplier quotes
Compare these vendor quotes for [Service/Deliverable]: [Quotes]. Create a table including price, scope of work, delivery timeline, contract duration, and notable observations. Afterwards, recommend one option and identify the two critical questions we should clarify before making a decision.
AI prompts for management
For management, AI acts primarily as a sparring partner: helping to structure decisions and challenge your own assumptions.
31. Decision template
I need to decide whether we should [Decision]. Background: [Context/Information]. Prepare a decision memo outlining the baseline situation, two to three options, pros and cons, estimated costs, and a recommendation. Clearly separate facts derived from my inputs from assumptions.
32. Challenging your own plan
Here is our plan for [Initiative/Project]: [Plan]. Adopt the role of a critical advisory board member. Identify the five weakest assumptions, what happens if they do not hold true, and which piece of information will show us most quickly whether we are on the right track.
33. Quarterly update for the team
Draft a quarterly leadership update to the team. Content: [Results, changes, goals]. The tone should be candid, even on challenging topics, acknowledge specific contributions from teams, and conclude with the priorities for the upcoming quarter. Maximum 300 words.
34. Competitor overview
Using web search, create an overview of our competitors [Names] in the [Market] market. Rely exclusively on publicly available information and cite the source for each claim. Compare their offerings, target audience, pricing strategy, and positioning in a table, and derive three strategic takeaways for us.
35. Outlining a presentation
I am delivering a presentation to [Audience] on the topic of [Topic], lasting [Minutes] minutes. Core message: [Message]. Create an outline with a maximum of [Number] slides. For each slide, provide an assertion-based headline, three bullet points, and a suggestion for a visual. The final slide must include a clear call to action for the audience.
AI prompts for IT and data
Even without a dedicated IT department, these prompts help you troubleshoot errors, write guides, and keep your data organized.
36. Understanding error messages
I am receiving this error message in [Software/Tool]: [Error Message]. Here is what I did beforehand: [Steps]. Explain in simple terms what this error means, identify the three most likely root causes, and provide a troubleshooting sequence to check them.
37. Guide for colleagues
Write a brief guide for colleagues without an IT background: [Task, e.g., set up two-factor authentication] in [Tool]. Number each step, keep each step to a single sentence, and add a "Troubleshooting" section at the end addressing the two most common issues.
38. Requirements for a new tool
We are looking for a tool for [Purpose]. Here are the requirements we have gathered: [List]. Categorize them into must-have, should-have, and nice-to-have. Add key missing considerations—such as data privacy, integrations, and pricing models—and use them to formulate ten targeted questions for vendor discovery calls.
39. Cleaning up a data list
Here is an anonymized extract from our customer database: [Data]. Identify duplicates, inconsistent formatting or spellings, and missing mandatory fields. List these issues and define a data hygiene rule for each to keep our database clean going forward.
40. Draft an AI usage policy
Draft an AI acceptable use policy for our company of [Number] employees. We use [Tools and subscription plans]. In at most one page, the policy should outline: approved tools, data that must never be input into AI tools, review responsibilities for AI-generated output, and points of contact for questions. Write in clear, plain language with no legalese.
Such a policy also helps with the training requirements under the EU AI Act. You can read about what is required there in our article on Mandatory AI training.
When a prompt is no longer enough
Prompts are perfect for one-off or occasional tasks. Once you find yourself copying the same prompt every week, it’s time to switch to a different tool:
| Scenario |
Better Suited |
Best For |
| You use the same prompt every week |
A custom skill or project with fixed instructions |
Individuals handling recurring tasks |
| The prompt requires templates, style guides, or reference files |
A custom skill that includes the reference files directly |
Teams with standardized formats |
| Multiple colleagues share the same workflow |
A shared team skill on a business plan |
Departments with common standards |
| The task needs to run automatically without manual chat input |
An automated workflow using tools like n8n or Make |
Cross-system processes |
At its core, a skill is a prompt with a memory: it contains instructions, examples, and files, and loads automatically when the task fits. Our articles on how this works can be found here: Claude Skills and ChatGPT Skills. For guidelines that should apply to every chat, often all you need are custom instructions.
If you want a task to run entirely without a chat - for example, automatically summarizing every new request and adding it to your CRM—a workflow is the better choice. Our article explains what that looks like AI automation explained simply.
Conclusion: Good prompts are specific, and the best ones become skills
Good AI prompts specify the task, source material, target audience, format, and constraints. With the 40 templates in this article, you have a starting point for every department that you just need to fill in with your own details. Always check the results, though, especially when it comes to numbers, names, and anything going to customers. And keep an eye on which prompts you use repeatedly: those are exactly the ones that are candidates for a skill or an automated workflow.
The best way to get started: Pick three prompts from this article that match your most frequent tasks and use them for a week. Afterward, tweak them to match your own phrasing and save the best versions in a place where the whole team can access them.
If you want to know which of your tasks are worth turning into a skill or an automation, get in touch with us for an AI consultation.
Ready to scale with AI?
Get a free strategy call and discover how AI can help you automate workflows, create better content, and scale your marketing.
Book a call
FAQ: Frequently asked questions about AI prompts
What is an AI prompt?
An AI prompt is the instruction you give to an AI assistant like ChatGPT or Claude. It describes the task and provides the necessary material. The more precise the prompt, the better the result.
How do I write a good AI prompt?
State the task, provide the material and target audience, and define the format and length. Add constraints, such as "do not invent any numbers." For important tasks, it’s worth adding a note that the AI should ask clarifying questions first.
Do these prompts work in ChatGPT, Claude, and Copilot?
Yes, the templates aren't tied to one specific tool. They work in ChatGPT, Claude, Gemini, Microsoft Copilot, and similar assistants. For prompt 34, you will need a tool with web search capabilities.
Are prompts in German worse than those in English?
No, current AI models understand German very well. How specifically you phrase your request is more important than the language itself. It’s best to write the prompt in the language you want the result in.
Where should I save my best prompts?
To start, a shared team document is enough. For prompts you use all the time, it’s better to save them as custom instructions, within a project, or as a skill. That way, you don't have to copy and paste them every time.
What’s the difference between a prompt and a skill?
A prompt is a single instruction that you enter from scratch each time. A skill permanently stores instructions, examples, and files, and loads automatically whenever a relevant task comes up.
.webp)
.webp)
.webp)
.webp)
