Prompt templates for business teams with AI, 26 to copy (2026)

A prompt template for business is a reusable instruction with fields in brackets that your team adapts to ask an AI for repeatable tasks. This guide gathers 26 to copy, by tool and area, and the method to structure your own. To roll it out across your team, see the online AI program from Founderz.

Add Founderz as a preferred source

What you’ll take from this

This guide gives you templates ready to customize and the keys to using them well in a team. Here is what it includes:

• 26 prompt templates to copy: marketing, sales, data analysis, documents, HR, finance and operations.
• Prompts built around what each tool does best: ChatGPT, Claude, Copilot and Gemini.
• The fields that make a business prompt work, with a weak example turned into a good one.
• How to store, version and measure templates in a team, small business or large company.
• Which data to keep out and what a person should always review.

Many teams reinvent the same AI request every week: one person writes a long prompt, another writes a different one for the same task and the results look nothing alike. A template fixes that. You change what is in brackets, copy the prompt and get a useful draft in minutes. Every one below defines a role, context, task, constraints and output format, and you can use them in a marketing, sales, HR, finance or operations team.

Templates serve companies and professionals alike, because they guide the AI in natural language and avoid generic answers. By predefining the specific task and the format, the AI model gets what it needs to give precise answers, and they work for content creation, internal processes or the business applications of each area.

ChatGPT prompts for business teams

ChatGPT is the most general option: it works for writing, for analyzing files you attach, such as a CSV, and for generating images. These templates use all three and are ordered by area.

ChatGPT prompts for marketing and content

Blog article outline

Act as a B2B content writer.
Context: my company sells [product or service] to [type of customer]. Their main problem is [problem].
Task: write the outline of an article about [topic] with a title, 5 H2 sections and one frequently asked question per section.
Constraints: [friendly / technical] tone, no promises we cannot keep and the brand name no more than twice.
Output: the outline as a numbered list and, at the end, three alternative headline ideas.

A 3-email marketing sequence

Act as an email marketing specialist.
Context: product [product], audience [segment], goal [welcome / reactivation / launch].
Task: write a sequence of 3 emails sent on day [0], day [3] and day [7].
Constraints: subject line under 50 characters, one message per email and a clear call to action. Tone: [brand tone].
Output: a table with the day, the subject line, the email text and the call to action.

Five campaign ideas with a metric

Act as the marketing director of a small business.
Context: brand [brand], budget of [amount], goal [acquisition / retention] and audience [audience].
Task: propose 5 different campaign ideas.
Constraints: none may need more than the stated budget and each must be launchable in [4 weeks].
Output: a table with idea, channel, format, effort (low, medium or high) and success metric.

ChatGPT prompts for sales and lead generation

In B2B, sales cycles are long and repetitive, which is why these prompts fit well. If you use a CRM such as HubSpot, generate the draft with the template and paste it into the contact record.

Qualify a lead

Act as a B2B sales manager.
Context: here is the data for a lead: [paste the CRM data here, leaving out personal information you do not need]. Our ideal customer is [ideal customer description].
Task: score the lead from 1 to 5 on fit, urgency and budget.
Constraints: use only the data I give you and mark anything missing as "no data" instead of assuming it.
Output: the score for each criterion with a one-sentence justification, the final score and the next action you recommend.

Follow-up sequence for a lead who does not reply

Act as a B2B sales executive.
Context: we sell [product]. The last contact with [name and job title] was on [date] about [topic] and they have not replied.
Task: write 4 follow-up messages spaced [3, 5, 7 and 10] days apart.
Constraints: professional and not pushy, each message adds something new (a fact, a case, a question) and none is over 90 words.
Output: the 4 messages, numbered, with a subject line and the day to send.

Draft a sales proposal

Act as a sales consultant preparing a B2B proposal.
Context: the client is [company, industry] and needs to solve [problem]. We have discussed [scope and timeline].
Task: write a draft proposal with a summary of the problem, the solution, the deliverables, the schedule and the next steps.
Constraints: do not invent prices or success stories; leave [price] and [client case] in brackets for me to complete.
Output: text with a heading for each section, around 400 words.

ChatGPT prompts for data analysis and creative work

Attach the file and the prompt tells it what to do with it. Check the important figures against your source before using them.

Analyze a sales CSV file

Act as a business data analyst.
Context: I am attaching a CSV with sales for [period] with the columns [date, customer, product, amount]. I removed personal data.
Task: tell me which product and which customer segment bring in the most revenue, and spot trends and outliers.
Constraints: use only the file, separate what you see in the data from what you interpret and do not round the figures.
Output: three findings with their figure, a summary table and two questions I should ask myself.

Segment customers from data

Act as a marketing analyst.
Context: I am attaching a customer table with [country, industry, size, purchases in the last 12 months].
Task: propose 3 to 4 segments based on the data and describe each one.
Constraints: each segment must have at least [20] customers and a clear rule for assigning any customer to it.
Output: a table with segment name, rule, number of customers, average value and one marketing action for each.

Image for a company post

Act as a graphic designer for a B2B brand.
Context: the post is about [topic] and is aimed at [audience] on [LinkedIn / the blog].
Task: generate a wide, horizontal image that illustrates [concept] in a [sober / flat illustration / photographic] style.
Constraints: no text inside the image, no third-party logos and free space on the left to add a headline.
Output: the image and an alternative version with a different composition.

Claude prompts for business teams

Claude stands out with long documents and careful writing, so these templates ask it to read a lot of material and write with judgment. Attach the complete files instead of summarizing them yourself.

Summarize a long contract or document

Act as an analyst who reviews documents carefully.
Context: I am attaching a [number]-page document: [contract / report / tender]. My company is [short description] and I am mainly worried about [deadlines / costs / liabilities].
Task: summarize the document and point out the clauses or sections that affect us most.
Constraints: cite the section or page for each point, do not interpret legal clauses and flag anything unclear.
Output: a 10-line summary, a risk table with section, risk and a question for the owner, and a list of open questions.

Write an internal AI use policy

Act as an operations manager writing an internal policy.
Context: we are a [industry] company with [number] people. We use generative AI for [tasks] and handle data about [customers / employees / nothing sensitive].
Task: write an AI use policy covering which data can be used, who reviews the outputs and what is prohibited.
Constraints: plain language for people without technical training, no more than 2 pages, and leave in brackets whatever legal or the data protection officer must validate.
Output: a document with numbered headings and a final checklist of 8 points.

Turn loose notes into a procedure

Act as a process consultant.
Context: below are the messy notes on how we do [process]: [text of the notes].
Task: turn them into a step-by-step procedure a new person can follow.
Constraints: do not add steps that are not in the notes; if information is missing, ask me questions before writing it.
Output: a numbered list with the owner, the tool and the result of each step, and a section of common mistakes.

B2B proposal from several sources

Act as a B2B proposal writer with editorial judgment.
Context: I am attaching the notes from the meeting with [client], their brief and our standard offer.
Task: write a [3-page] proposal that answers what the client asked for, in their own words where possible.
Constraints: do not contradict the standard offer, do not promise deadlines that are not in the documents and mark [in brackets] any missing data.
Output: a proposal with an executive summary, approach, schedule and terms, and at the end a list of what I must confirm with the client.

Copilot prompts for business teams

Copilot works inside Word, Excel, Outlook and Teams with the files, emails and meetings you already have access to. If your team lives in Microsoft 365, it is the most direct way to apply the templates without leaving the tools you use every day. Name the file or the meeting in the prompt.

Summarize a Teams meeting and pull out the tasks

Context: the Teams meeting [meeting name] on [date] with the [area] team.
Task: summarize the meeting and extract the decisions, the tasks and the open questions.
Constraints: use only what was said in the meeting, assign each task to the person who accepted it and mark "no owner" when nobody took it on.
Output: three sections (decisions, tasks, questions) and a task table with owner and date.

Reply to a customer email in Outlook

Act as an account manager.
Context: summarize the email thread with [customer] about [subject] and keep in mind that our goal is to [keep the deadline / renegotiate the scope].
Task: write a reply to their last message.
Constraints: professional and friendly tone, no more than 150 words, answer every question they asked and do not commit to new dates.
Output: the thread summary in 3 lines and the draft reply with a subject line.

Draft a report in Word from a file

Act as an analyst preparing a report for [management].
Context: use the document [file name] as the main source and the workbook [Excel name] for the figures.
Task: write a [2-page] draft report in Word about [topic].
Constraints: do not include data that is not in those files, say which file each figure comes from and write "to be completed" where information is missing.
Output: title, executive summary, 3 sections with subheadings and a list of next steps.

Analyze an Excel table

Act as a data analyst.
Context: the table [Orders] in the open workbook has [Sales rep, Date, Amount, Status].
Task: calculate the amount per sales rep and per month, and tell me who is below the average.
Constraints: use formulas instead of typed values and do not change the original data.
Output: a summary table on a new sheet, the chart that best shows the comparison and one sentence with the conclusion.

Gemini prompts for business teams

Gemini fits if your team works with Google Workspace: Docs, Gmail and Sheets. Mention the document, the thread or the sheet by name.

Draft in Google Docs from your notes

Act as an internal communications manager.
Context: use the document [document name] with the notes for [project] and the style of the document [name of an example].
Task: write in Docs an announcement for [the whole team] about [news].
Constraints: 200 words at most, [friendly] tone, no jargon and a single action requested from the readers.
Output: title, text and a subject line to send it by email.

Summarize a Gmail thread and suggest a reply

Context: the Gmail thread with [customer or supplier] about [subject].
Task: summarize in 5 lines what has been agreed and what is still unresolved, and propose a reply.
Constraints: highlight dates and amounts, do not invent commitments and keep the tone of the earlier messages.
Output: the summary, a list of points to resolve and the draft reply.

Classify customer comments in Sheets

Act as a customer experience analyst.
Context: in the sheet [name] the column [Comment] contains customer opinions.
Task: classify each comment by topic ([price, support, product, delivery, other]) and by sentiment (positive, neutral, negative).
Constraints: write the result in two new columns without touching the original and flag doubtful comments.
Output: the two columns and a summary table with the count by topic and sentiment.

More prompt templates by area: HR, finance and operations

These six work in any tool. They are tasks with a clear output format, the ones that repeat most and where the savings show up first.

A clear, unbiased job posting (HR)

Act as a recruiting manager.
Context: the company is [industry, size], the role is [role] and it reports to [area]. The main tasks are [tasks] and the must-have requirements are [requirements].
Task: write a job posting.
Constraints: separate must-have from nice-to-have requirements, avoid language biased by gender or age and state the salary range [range].
Output: title, 3-line summary, tasks, requirements, what we offer and how to apply.

Structured interview script (HR)

Act as a competency-based hiring specialist.
Context: we are interviewing for [role] and the key competencies are [competency 1, 2 and 3].
Task: prepare a [45-minute] interview with the same questions for every candidate.
Constraints: questions about real past situations, nothing about personal life and a 1 to 5 rating scale with criteria.
Output: a table with competency, question, sign of a good answer and red flag.

Monthly variance commentary (finance)

Act as a financial controller.
Context: I am pasting the [month] table with budget and actual by line: [table]. The main lines are [lines].
Task: explain the 3 most relevant variances and their possible causes.
Constraints: calculate each variance in amount and percentage, and do not state causes as facts: present them as hypotheses to confirm.
Output: one paragraph per variance with the figures and a question for the line owner.

Monthly summary for management (finance)

Act as a financial analyst writing for [the management committee].
Context: data for [month]: [revenue, costs, margin, cash].
Task: write a 5-sentence summary with the result, the best, the worst and a decision management should make.
Constraints: only figures I give you, compared with the previous month and with the budget.
Output: text ready to paste into an email and a table of 4 indicators.

Checklist for a repeated process (operations)

Act as an operations manager.
Context: our [area] team repeats the process [process] every [frequency] and sometimes steps get forgotten.
Task: create a checklist that can be used on every run.
Constraints: 12 points at most, each checkable with yes or no and grouped into before, during and after.
Output: the checklist as a list and the 3 mistakes it is most likely to prevent.

Meeting minutes with tasks (operations)

Act as a project coordinator.
Context: I am pasting the notes from the [project] meeting on [date]: [text of the notes].
Task: write minutes with the decisions and the tasks.
Constraints: do not add anything that is not in the notes and write "to be confirmed" for owners or deadlines that were not mentioned.
Output: minutes under 250 words and a task table with owner, deadline and status.

How do you create your own prompt templates step by step?

Every template above follows the same structure, a business version of prompt engineering. If you want the conceptual framework, look up what prompt engineering means for a company. These are the fields, in this order:

Prompt wording improves with trial and error: send a first draft, see what fails, refine the prompt and test again. For complex tasks, give the model the target audience, the specific needs of the ideal customer and the datasets it should use, and ask for a clear output. That way you get clear prompts and make the most of each tool.

Five tips for your team to use them

1. Start from a task you already repeat every week: that is where the savings show most.
2. Put in brackets only what changes on each use and keep everything else fixed.
3. Ask for a format you can paste straight into an email, a CRM or a spreadsheet.
4. Tell the AI what to do when data is missing, for example “mark as no data” instead of assuming it.
5. Test the template with two different people and keep only the version that gives similar results.

Knowing how to write a good prompt comes from comparing versions. To generate content or write an email, you first see a version that works to copy and paste, and then another that gets the most out of your data.

Example: from a weak prompt to a good template

Weak prompt, which leaves almost everything to chance:

Write a sales email.

The same request with the fields, turned into a reusable template:

Act as a B2B sales executive.
Context: we sell [product] to [type of customer]. I spoke with [name and job title] on [date] and they agreed to [agreement].
Task: write a follow-up email proposing [next step].
Constraints: professional and friendly tone, under 120 words and no discounts or deadlines that do not appear above.
Output: subject line and email body, plus a shorter version to send as a message.

A consultant, a freelancer or the owners of a small business can start with three templates and a shared document. A good prompt, once validated, becomes a framework others can repeat and optimize. The checklist below tells you whether it is complete.

Checklist for a well-built template

• Role: does it say which profile the AI should answer from?
• Context: does it include the data and the industry needed?
• Task: is it clear what result you want?
• Constraints: length, tone and things to avoid?
• Output format: have you defined how you want the result?
• Sample data: is there a case that guides the answer?
• Success criterion: can you tell a good output from a bad one?

If a prompt meets all seven points, it is a solid template. If it does not convince, that is usually why. Iterate on each version with one concrete change, such as “make it shorter” or “add a real example”.

How the prompt changes in ChatGPT, Claude, Copilot and Gemini

• ChatGPT: attach the file or paste the data and ask for a specific format. It works for writing, analysis and image generation.
• Claude: give it the complete document and ask it to cite the section for each point. It is useful for careful reading and writing.
• Copilot: name the Microsoft 365 file, email or meeting and ask for the result in the app where you work.
• Gemini: point to the Google Workspace document, thread or sheet and ask for the result in that same app.

Comparison of recommended AI models for your templates

Plans and features change often: check the current terms of each tool before deciding.

How to build templates into your team workflow: automation and good practices

A template only adds value if your team can find it and uses it. The challenge is not creating good prompts but governing them: where they are stored, who updates them and how they connect with the rest of your tools.

• A shared repository: a document, a wiki or a folder the whole team can access, organized by area.
• Versions: note what changed and why, so you do not lose the one that worked.
• CRM and ERP: generate the draft with the template and paste it into the customer or process record.
• Automation: when a task repeats a lot, automate the data flow into the prompt.
• Internal policies: which data can be attached, who reviews the outputs and how they are approved.
• AI agents: for complete flows where the AI chains several tasks with human supervision.

A small business can start with one shared document and three templates. A large company needs formal versioning, permissions and training by area. These AI prompts for marketing briefs and calendars show how to take the same method into a specific workflow.

How do you measure whether your templates work?

You measure the result of a template in three ways: time saved, quality of the output and consistency between people.

• Time saved per task: compare the time before and after using the template.
• Output quality: how often you use it with almost no editing.
• Consistency: how similar the outputs are between different people.
• Adoption: how many templates the team uses regularly.

To report to management, document an internal success story with your own figures. Treat third-party headlines promising to “save 5 hours a day” with caution: they can guide you, but do not take them as your own data until you measure them in your context. And some decisions still need human judgment: a delicate sales decision, assessing a person in HR or a financial interpretation with nuance. The AI prepares the draft and you decide.

Security and privacy with internal data

• Anonymize sensitive data: remove names, identifiers and personal data when they are not needed.
• Check the provider policy: find out what it does with the data you send and which plan your company uses.
• Keep human oversight: no output is published or sent to a customer without review.
• Limit what goes in: the less sensitive data you attach, the lower the risk.

Frequently asked questions about prompt templates for business

These are the most common questions about how to use prompt templates in a company and which tool to choose.

What is a prompt template?

It is a reusable instruction with fields in brackets that you change each time. The fixed parts (role, task, format) are written once and you fill in the variable parts (customer, topic, data), so anyone on the team gets comparable results.

What are good examples of prompt templates for ChatGPT?

The ones that ask for a specific business task with a role, context and format: an article outline, an email sequence, a lead qualification or the analysis of a sales CSV. You get 7 ready to copy in the ChatGPT section and 6 more by area.

How do I create a prompt template for my team?

Start from a task that repeats, write a prompt that works with a role, context, task, constraints and output format, and replace the data from that case with brackets. Test it with two people, fix what fails and save it with a clear name.

Do the same templates work in Claude, Copilot and Gemini?

The structure does, but each tool is used differently. Claude does well with long documents, Copilot works with your files in Word, Excel, Outlook and Teams, and Gemini with Google Workspace. Change what each one can read, not the method.

Where do I store and version prompt templates?

In a repository your team shares, such as a document, a spreadsheet or an internal wiki. Organize them by area, note in each version what changed and why, and name one person to maintain them.

How do prompt templates fit with a CRM?

As an intermediate step: you generate the draft with the template and paste it into the contact record, in HubSpot for example, to keep the history in one place. If the task repeats a lot, you can automate the data flow between the CRM and the prompt.

How do I know whether a template works?

With four signals: time saved per task, how often you use the output with almost no editing, consistency between people and the number of templates the team actually uses. Measure in your own context and distrust third-party figures with no method behind them.

Is it safe to attach internal data to ChatGPT?

It depends on the data and the plan you use. Anonymize sensitive information, share only what is needed, check the provider’s data policy and your company rules, and do not publish any output without human review.

Are there free prompt templates to start with today?

Yes. Copy any template on this page, change the brackets and try it in the free version of ChatGPT, Claude or Gemini. If you prefer to build one from your own options, the free prompt template tool linked on this page prepares it for you.

Your next step: bring the templates to your whole team

Pick three tasks your team repeats every week, adapt one template for each and store them in a shared place. After a week of use you will know which ones are worth keeping.

To learn to apply AI with a method in your role, the online AI program from Founderz gives you the foundation with practical cases and a joint certification with Microsoft.