Okonkwo runs a six-person landscaping company outside Columbus. He started using an AI writing tool eight months ago, mostly for customer emails. Early on he would type something like "write me an email about a quote" and get back a generic blob that sounded nothing like him and missed half the details, so he rewrote most of it. It was faster than starting from scratch, but not by much. Then a friend showed him a simple trick: describe the role you want the AI to play, the task, and the constraints, before you give it any content. Okonkwo tried it. His next quote follow-up took four minutes instead of twenty, needed no rewrites, and the customer replied the same day. The output changed because the input structure changed.

Why Patterns Beat One-Off Prompts

A prompt is a single instruction for a single task. A prompt pattern is a reusable template that covers a whole category of tasks. The difference is a recipe versus a specific meal: you write the recipe once and use it hundreds of times. Software developers have worked this way for decades, reaching for established design patterns rather than solving each common problem from scratch, and the same logic applies to prompts.

Five patterns cover the vast majority of what people need from AI tools at work. Each one specifies a role for the AI, a task type, and a set of constraints. Once you have a pattern that works, you save it, in a document, a notes app, or a text expander, and reuse it with only the specific details changed. Each pattern below appears twice: once as a fill-in template, once as a worked example.

Those examples come at two different scales, and it is worth keeping them apart. Okonkwo's scenes are a six-person owner-operated business where the constraint is his own time. The worked examples that follow each pattern come from a larger setting, a forty-person software company, where budgets run to six figures and the constraint is coordination. The patterns are identical at both scales. The numbers are not, so do not carry a figure from one to the other.

The Analyst Pattern

When to use it: examining data, extracting insights, evaluating options, making recommendations. Best for: metrics review, customer feedback analysis, vendor evaluation, competitive research, financial analysis, performance assessment. Use it when you need to make sense of information you already have: a pile of customer reviews, a set of sales numbers, a list of vendor proposals.

Template: "You are an expert analyst specializing in [YOUR FIELD]. [CONTEXT ABOUT YOUR SITUATION] Analyze the following [DATA/INFORMATION]: [INSERT DATA OR COPY-PASTE CONTENT] Specifically, identify: [3-5 SPECIFIC INSIGHTS YOU WANT]. Format your analysis as: [FORMAT]. Keep to [LENGTH LIMIT]. Assume [AUDIENCE LEVEL]."

A short version works for a small business. Okonkwo uses this shape: "You are a business analyst. Review the following [data/feedback/proposals] and give me: (1) the three most important patterns or insights, (2) one specific recommendation, and (3) any gaps in the information I should fill before deciding." He ran it over 40 Google reviews of his company. He pasted them in and got a clear summary: customers loved his crews but complained about quote response times. That took ten minutes instead of an hour of manual reading. More importantly, the pattern forced a recommendation rather than just a summary. He now follows up on quotes within 24 hours, and his review scores improved.

At larger scale the same pattern carries far more context. Here is the full version, for a software company analyzing churn risk:

"You are an expert in customer experience and user research. Our company is a B2B SaaS project management tool with 200 mid-market customers. We're analyzing churn risk. Analyze this customer feedback from our recent survey: [PASTE 20 FEEDBACK RESPONSES HERE] Specifically identify: 1) The top 3 pain points mentioned, 2) Which customer segments express which problems most, 3) Quick wins we could address immediately, 4) Structural issues requiring product changes. Format as: Pain Point (with quote examples), Affected Segments, Severity (High/Medium/Low), Recommended Action. Keep under 500 words. Audience: Our product team, non-technical but analytical."

The Analyst pattern works because it constrains the AI to examine rather than opine. It demands reasoning from the data in front of it instead of new ideas, which is exactly what you want when you need facts extracted rather than suggestions generated. Keep the ask tight. Do not request a list of everything; ask for the three most important things, because a length constraint forces prioritization, and the prioritization is the part you cannot easily do yourself.

The Strategist Pattern

When to use it: big-picture thinking, growth opportunities, competitive positioning, business direction. Best for: annual planning, market entry strategy, product positioning, pricing strategy, go-to-market planning, competitive response. Reach for it when you are making a significant decision and need to think it through before committing: pricing a new service, deciding whether to hire or subcontract, evaluating a new market.

Template: "You are a strategic business consultant with [RELEVANT EXPERIENCE]. Your goal is to help [YOUR COMPANY] achieve [STRATEGIC GOAL]. Here's our context: [COMPANY OVERVIEW], [MARKET POSITION], [KEY CONSTRAINTS], [TIMEFRAME]. Develop a strategic recommendation addressing [SPECIFIC CHALLENGE]. Your recommendation should: [CRITERIA 1], [CRITERIA 2], [CRITERIA 3]. Structure your response as: Executive Summary (150 words), Strategic Pillars (3-4 main approaches), Implementation Timeline, Key Metrics to Track, Risks and Mitigation."

Okonkwo's version of the same pattern asks for less and withholds one thing deliberately: "You are a business strategist. I'm considering [decision]. My business is [brief description]. My main constraint is [budget/time/capacity]. Give me: (1) the strongest argument for doing this, (2) the strongest argument against, and (3) what I'd need to know before committing." Notice what it does not ask for, which is a recommendation. That is intentional. At the strategic level you want the AI to surface considerations, not make calls on your behalf, because you hold context it does not: your gut feel about a supplier, a long-standing customer relationship, a cash position you have not shared.

The fuller template earns its length when the decision is big enough to justify it. A worked example from the larger scale, a company considering a move into a new customer segment:

"You are a go-to-market strategist with experience helping B2B SaaS companies enter new markets. Your goal is to help us expand our project management software to a new customer segment. Our context: We're a 40-person SaaS company with $3.2M ARR serving mid-market companies (50-500 people). Our product is strong in enterprise-grade features. We currently focus on operations and finance teams. We want to expand to marketing departments but don't know if product or positioning changes are needed. We have $100k to invest in this expansion over 12 months. Develop a market entry strategy for the marketing department segment. Your strategy should: Address whether we need product changes, clarify positioning differences, identify which marketing roles are our ideal customer, recommend go-to-market approach (sales-led or self-serve). Structure as: Executive Summary (150 words), Market Opportunity Analysis, Positioning Recommendation, Ideal Customer Profile for Marketing, Recommended Go-to-Market Approach, 12-Month Implementation Timeline, Key Success Metrics."

The Strategist pattern forces big-picture thinking and resists the urge to jump to tactics before strategy is settled. It requires reasoning about competitive positioning and business direction rather than features and functions. At Okonkwo's scale, the useful threshold is modest: run this pattern before spending more than $500 or committing more than a week of your own time. Ten minutes of structured thinking regularly surfaces a missing piece you would otherwise have discovered after the fact.

The Editor Pattern

When to use it: improving, refining, and polishing existing content. Best for: email refinement, proposal editing, copy improvement, tone adjustment, clarity checking, messaging alignment. Use it when you have already written something and want it to be better: a proposal, a contract email, a reply to an angry customer, a job posting.

Template: "You are an expert editor specializing in [TYPE OF CONTENT] for [AUDIENCE TYPE]. Your job is to make this content more [DESIRED QUALITY]. Here's the content I've drafted: [INSERT YOUR DRAFT] Edit this for: [SPECIFIC IMPROVEMENTS, for example clarity, conciseness, persuasiveness, tone]. Specifically, [CONSTRAINTS, for example keep it under 150 words, make it more casual, add social proof]. Return the improved version and note the 3 most important changes you made."

Okonkwo's variant adds one clause that is worth copying: "You are an editor. Improve the following text for [audience: customer / job applicant / vendor]. Goals: [clear / professional / friendly / persuasive]. Do not change the meaning or any factual details. Show me the revised version and briefly explain what you changed and why." The explanation clause is the important part. It tells you whether the AI understood your intent, and if its explanation is off, the revision probably is too. You can course-correct in one turn rather than puzzling over why the output feels wrong.

A worked example at the larger scale, editing a sales email:

"You are an expert editor specializing in B2B SaaS emails for busy operations managers. Your job is to make this email more compelling and action-oriented. Here's the email I've drafted: [PASTE YOUR EMAIL DRAFT] Edit this for: clarity (make the ask obvious), urgency (add light urgency without pressure), personalization (make it feel less generic), and conciseness (cut anything that doesn't drive the ask). Specifically, keep it under 200 words, maintain a professional-but-friendly tone, and include a clear single call-to-action. Return the improved version and note the 3 most important changes you made."

The Editor pattern is what you want when you have draft content that is not quite right. It never asks the AI to write from scratch; it refines and improves your work, which produces better results and keeps the voice yours. Okonkwo uses it almost daily. He drafts in his own voice, fast and unpolished, then runs the Editor pattern to tighten the language, and the result sounds like him, only cleaner. That matters for a small business, where your emails should sound like you rather than like a corporate template.

The Teacher Pattern

When to use it: creating educational content, explaining concepts, training your team, clarifying complex topics. Best for: onboarding guides, training materials, explanation emails, documentation, internal education, knowledge transfer. Use it when you need to explain something to an employee, a customer, or a vendor and you are not sure how to frame it simply.

Template: "You are an expert educator teaching [TOPIC] to [AUDIENCE]. Your goal is to make this concept crystal clear, even if they have no background in [FIELD]. Explain [SPECIFIC CONCEPT/TOPIC] in a way that: [CRITERIA 1, for example uses real business examples], [CRITERIA 2, for example avoids jargon], [CRITERIA 3, for example includes step-by-step process]. Format as: Overview (1 paragraph), Why This Matters (1 paragraph), Key Concepts (3-4 bullet points with definitions), Real Example (with walkthrough), Common Mistakes to Avoid (3 bullets), Action Step (what to do now)."

Okonkwo's version compresses that into three lines: "You are a teacher explaining to [audience who has no background in this topic]. Explain [concept] in plain language. Use one concrete example from a [similar industry]. Keep it under 200 words." He used it to create a one-page handout explaining to new crew members why they should take before-and-after photos of every job. He had tried explaining it verbally three times. The handout, generated in eight minutes and edited in five, did it in a way that stuck. The word limit is the discipline: short explanations get read, long ones get skimmed and forgotten.

The longer format shape suits material with real structure behind it. A worked example, onboarding a new sales hire onto a commission plan:

"You are an expert onboarding coach for sales teams. Your goal is to help a new team member understand our sales process and compensation structure. Explain our commission structure: 3% commission on total deal value for deals under $50k, 4% for deals $50k-$150k, 5% for deals over $150k. Accelerator: 6% on any deal revenue over quarterly target. Annual salary $80k plus commissions. Our average deal is $25k, typical sales cycle is 6 weeks, and reps close 20-25% of qualified demos. Format as: Overview (1 paragraph), Why It Matters (1 paragraph), Commission Formula (clear breakdown with example math), Realistic Earnings Expectation, Common Questions, Action Steps (understand your current pipeline, calculate potential earnings)."

The Teacher pattern forces clarity. It stops you assuming the audience understands something they do not, and it requires a step-by-step breakdown with concrete examples rather than a definition. Note how much of that example is numbers: when the thing you are explaining has rules, feed in every rule, because an explanation built on half of them is worse than none.

The Devil's Advocate Pattern

When to use it: testing ideas, finding holes in plans, anticipating objections, stress-testing strategies. Best for: pre-mortem analysis, objection handling, risk assessment, decision validation, competitive threat assessment, market entry risk analysis. Use it before you commit to a plan: a new service package, a significant purchase, hiring a new employee.

Template: "You are a critical thinking expert and devil's advocate. Your job is to find the holes and risks in our thinking. Be direct and challenging. Here's what we're proposing: [DESCRIBE YOUR PLAN/IDEA] Play devil's advocate by: 1) Identify the 3 biggest assumptions we're making, 2) Argue why this plan might fail spectacularly, 3) Identify which customer segments or scenarios would reject this, 4) Highlight what competitive threats could undermine this. Format as: Biggest Assumptions (numbered list), Failure Scenarios (3 ways this could fail), Dissatisfied Customers (which segments might hate this), Competitive Threats (how competitors could counter this), Overall Risk Assessment (High/Medium/Low)."

Okonkwo's short form: "You are a skeptic. I'm planning to [action]. Argue against this plan. What assumptions am I making that could be wrong? What could go wrong in the first 90 days? What would I regret not having thought about?" This is the most underused of the five patterns. Most people use AI to confirm ideas rather than challenge them, and this pattern forces the tool to do the opposite. It finds the holes.

When Okonkwo considered adding a snow removal service last winter, he ran it. The AI raised three things he had not fully thought through: equipment maintenance costs in heavy-use months, liability if a crew slips on ice, and whether his existing customers were even in areas that got enough snow to justify a truck. He scaled the plan back, rented equipment instead of buying it, and limited the service to a trial run with five accounts. That trial cost $2,000 instead of $18,000, and it worked well enough to justify a permanent service the following season.

A worked example at the larger scale, stress-testing a product launch:

"You are a critical thinking expert and devil's advocate. Your job is to find the holes and risks in our product launch strategy. Here's what we're proposing: We're launching a new 'automation' feature for our project management software next month. We're adding job scheduling, workflow automation, and integration with Zapier. We think this will let us charge $100/month more (moving from $300 to $400/month). We're targeting mid-market operations teams. We haven't conducted customer research on this specific feature. Play devil's advocate by: 1) Identify the 3 biggest assumptions, 2) Argue why customers will reject this, 3) Identify which segments would have strong objections, 4) Highlight what competitors could do to undermine this. Format as: Critical Assumptions, Failure Scenarios (3 specific ways this launches and falls flat), Dissatisfied Customer Segments, Competitive Threat Response, Overall Risk Assessment with confidence level."

The Devil's Advocate pattern is essential before major decisions. It prevents groupthink and forces you to consider what could go wrong instead of assuming success. Its output is uncomfortable by design, which is why so few people run it and why it should precede any decision large enough to hurt.

Choosing and Combining Patterns

The five patterns cover the majority of business tasks, and the quickest way to pick one is to ask what kind of output you need rather than what topic you are working on.

The real power comes from combining them. You might use the Strategist pattern to develop a strategy, then the Devil's Advocate pattern to test it, then the Teacher pattern to explain the result to your team. Each stage takes the previous output as its input, and each stage does one job well. The chain also exposes a weak plan early: if the Devil's Advocate demolishes what the Strategist produced, the problem is the plan, not the prompt.

Building Your Pattern Library

Save each pattern as a named template in whatever system you already use, whether that is a documents app, a notes app, or a plain text file. Label them clearly: Analyst, Strategist, Editor, Teacher, Devil's Advocate. When a task comes up, pick the pattern that fits the task type, paste in your specific details, and run it. The naming matters, because a pattern you cannot find quickly is one you will rewrite from scratch.

Over time you will develop variations: an Editor pattern tuned for angry customer replies, an Analyst pattern built specifically for review analysis, a Teacher pattern for your employee handbook. That library is yours, and it gets more valuable every time you use it. The investment to build it is about an hour, enough to set up the five patterns and test each one against a real task from your own business. The return is that every AI-assisted task after that runs faster and produces better output than starting from scratch.

Anti-Patterns to Avoid

Patterns fail in predictable ways, and most of the failures come from using the right pattern for the wrong kind of task.

• Asking the Strategist for the decision. The pattern is built to surface considerations. You hold the context it does not have, so the call stays yours.
• Using the Analyst when you want ideas. It is constrained to examine rather than opine. If you need new options generated, this is the wrong pattern.
• Letting the Editor rewrite from scratch. The pattern exists to keep the voice yours. Drop the instruction not to change meaning or facts and you get a different document rather than a better one.
• Dropping the explanation clause. Without it you cannot tell whether the AI understood your intent, and you end up re-reading the revision line by line.
• Removing the length limits. Both the Analyst and the Teacher depend on a constraint to force prioritization. Unlimited output is unprioritized output.
• Running the Devil's Advocate after you have committed. Used as reassurance rather than as a test, it is theatre. It has to run while the plan can still change.
• Mixing scales. A six-figure expansion budget and a five-hundred-dollar decision threshold belong to different businesses. Reuse the pattern, not the other business's numbers.
• Keeping the library in your head. An unsaved pattern is a pattern you rebuild every time, which is exactly the cost patterns exist to remove.

Practice Prompts

Work through these with real material from your own business, using the templates above as written. These exercises tell you which pattern to reach for and what to notice in the output.

• Analyst on your own reviews. Paste your recent customer reviews or survey responses into the short Analyst form. Check that the output ends in a recommendation and a list of gaps, not just a summary.
• Analyst, full form. Re-run the same data through the long template with a format specification and a length limit. Compare which version you would actually act on.
• Strategist without a recommendation. Take a decision you are facing and run the short Strategist form. Resist the urge to add "what should I do?" and see whether the two arguments settle it anyway.
• Editor with and without the explanation clause. Run the same draft twice, once asking for the reasoning behind the changes and once not, and note how much faster you can accept or reject the version that explains itself.
• Teacher against something you have explained three times. Pick the thing your team keeps getting wrong, set a 200-word limit, and turn the output into a one-page handout.
• Devil's Advocate on a live plan. Run it on something you have not yet committed to. Write down which of its objections you had genuinely not considered.
• Chain three patterns. Run the Strategist on a decision, feed its output into the Devil's Advocate, then use the Teacher pattern to write the version you would give your team.

Reflection

Think about the last few things you asked an AI tool to do and ask which of the five patterns each one was reaching for, even if you did not name it at the time. Most people find their requests cluster into two patterns and ignore the rest. The neglected ones are usually the most valuable, because they represent thinking you were not doing before, rather than typing you were already doing.

The Devil's Advocate is the clearest case. Almost nobody runs it, because asking a tool to attack your own plan invites bad news, and it usually delivers. Okonkwo's snow removal trial is the argument for it: the pattern did not stop him doing the thing, it stopped him doing the expensive version. Ask what you committed to last year that would have survived ten minutes of that question.

Glossary

• Prompt pattern: A reusable template covering a category of tasks, specifying a role, a task type, and constraints.
• Role assignment: Telling the AI what perspective to take, which shapes what it weighs and how it reasons.
• Bracketed placeholder: A field such as [YOUR FIELD] or [PASTE] that you replace with your own detail before running the prompt.
• Explanation clause: An instruction asking the AI to say what it changed and why, used to check that it understood your intent.
• Pre-mortem: Imagining a plan has already failed and reasoning backward to the causes, which is the Devil's Advocate pattern's core move.
• Pattern chaining: Feeding the output of one pattern into another, for instance Strategist into Devil's Advocate into Teacher.
• Pattern library: Your saved, named collection of patterns and the variations you have tuned for recurring tasks.

Related Lessons

• The Anatomy of an Effective Prompt breaks down the five components every one of these patterns is assembled from.
• Common Prompting Mistakes and How to Fix Them covers the failures these patterns are designed to prevent.
• Building Your Personal Prompt Library goes further on organizing, versioning, and improving the library this lesson starts.
• Advanced Prompt Structures and Chaining develops the combination technique into multi-stage workflows.
• Prompt Templates for Daily Business Operations supplies ready-made prompts for recurring tasks alongside these five patterns.

Closing

Five proven patterns, Analyst, Strategist, Editor, Teacher, and Devil's Advocate, cover the vast majority of business tasks. Rather than writing new prompts from scratch, use these as templates, customize the context and task for your own situation, and you will get consistent output without rebuilding the structure every time. Build a library of them with your own variations and you will write prompts faster while the quality holds steady rather than drifting with your mood and your available time.

The patterns scale in both directions, which is the point of seeing them at two sizes. The same Analyst structure that turns forty reviews into one decision for a six-person landscaping company turns a survey into a churn diagnosis for a forty-person software company. What changes is how much context you feed in and how tightly you constrain the output. What does not change is that you name a role, describe a task, and set the limits before handing over any content.

Key Takeaways

• A prompt is a single instruction; a pattern is a reusable template for a whole category of tasks. Write the recipe once, use it hundreds of times.
• Five patterns cover most business work: Analyst, Strategist, Editor, Teacher, Devil's Advocate.
• The Analyst extracts signal from noise. Ask for three insights, one recommendation, and any gaps, and constrain the length so it prioritizes.
• The Strategist surfaces what you have not considered. Ask for the best argument for, the best argument against, and what you would need to know, then make the call yourself.
• The Editor improves without replacing your voice. Tell it not to change meaning or facts, and keep the explanation clause so you can tell whether it understood you.
• The Teacher simplifies explanation. Give it an audience, a concept, one concrete example, and a word limit; short explanations get read.
• The Devil's Advocate finds holes before they cost money, and it is the most underused of the five because its output is unwelcome by design.
• Patterns combine: Strategist to develop, Devil's Advocate to test, Teacher to explain.
• Keep the numbers with the scale they came from. Okonkwo's $500 decision threshold and a $100k expansion budget belong to different businesses.
• Save your patterns in a named library. About an hour of setup covers all five, and it pays for itself on the first reuse.

Frequently Asked Questions

What are prompt patterns and why do they matter?

Prompt patterns are reusable templates that specify what role the AI should take, what task to perform, and how to approach it. They matter because they remove the need to design prompts from scratch every time. Once you have a pattern that works well, you can reuse it hundreds of times with only minor modifications for different contexts, which saves time and produces consistent quality.

What is the Analyst pattern used for?

The Analyst pattern is used for examining data, extracting insights, evaluating options, and making recommendations. It suits reviewing metrics, analyzing customer feedback, evaluating vendor options, competitive research, or any task where you need objective analysis of existing information rather than creative generation. The pattern constrains the AI to examine and reason rather than opine.

When should I use the Strategist pattern?

Use the Strategist pattern when you need big-picture thinking about direction, growth opportunities, competitive positioning, or business challenges. It suits annual planning, market entry strategy, product positioning, and major business decisions. The pattern resists jumping to tactics and forces reasoning about competitive context and business direction before implementation detail.

What is the Editor pattern for?

The Editor pattern is used for improving, refining, and polishing existing content. Use it to edit emails before sending, improve proposals, refine copy, check for clarity, ensure the tone is appropriate, and make content more compelling. The Editor pattern always starts from content you have drafted, which is what makes it good at refining your work without replacing your voice.

How are prompt patterns different from individual prompts?

A prompt pattern is a template you reuse; an individual prompt is written for one specific task. Patterns save time because you build the structure once, then use it repeatedly by changing only the context and the task. One well-designed pattern can be used hundreds of times, and patterns also keep you consistent in how you approach similar kinds of work.