What Is the CRIT Framework? A Better Way to Use AI for Business
June 4, 2026Fifty-eight percent of small businesses now use generative AI. That number jumped from 40 percent in just one year. But here’s the part nobody talks about: most of those business owners type something vague into ChatGPT, get a generic wall of text back, and quietly decide AI isn’t that useful. The tool isn’t the problem. The way you talk to it is.
Geoff Woods wrote about this in his book The AI-Driven Leader. His argument is simple. Most people treat AI like a search engine. They throw a question at it and hope for the best. That’s like hiring a consultant, giving them zero context about your business, and then complaining that their advice is generic. Of course it is. You gave them nothing to work with.
Woods built a framework called CRIT to fix this. Four letters. Four steps. It takes about 60 seconds to learn, and it changes what AI gives you from forgettable filler to something you’d actually use. If you run a business and you’ve been underwhelmed by AI so far, this is probably why.
What You’ll Learn
- Why Do Most AI Prompts Give You Generic Junk?
- What Is the CRIT Framework?
- How Context Changes Everything
- Why Assigning a Role Matters More Than You Think
- What Makes the Interview Step So Powerful?
- Defining the Task So AI Actually Delivers
- What Does CRIT Look Like in Practice?
- Where Should You Start?
Why Do Most AI Prompts Give You Generic Junk?
The average AI prompt from a small business owner looks something like this: “Write me a blog post about marketing.” Or “Give me ideas for social media.” Or “Help me with my website copy.”
These prompts aren’t wrong. They’re just empty. There’s no information about who you are, who your customers are, what you’ve already tried, or what you’re actually trying to accomplish. The AI has nothing specific to work with, so it gives you the most average possible answer. It’s doing exactly what you asked. You just didn’t ask for much.
Think about it from the other direction. If someone walked into your business and said “help me,” your first question would be “with what?” Your second question would be “what have you already tried?” Your third would be “what does success look like for you?” You’d need answers before you could give useful advice. AI works the same way. The difference between a useless response and a genuinely helpful one almost always comes down to what you put into the prompt, not which AI tool you’re using.
This is the gap that Geoff Woods identified. He spent years as Chief Growth Officer at Jindal Steel & Power, where his guidance helped grow their market cap from $750 million to over $12 billion. When he started applying AI to business strategy, he realized the same principle applied: better inputs create better outputs. So he built a system for it.
What Is the CRIT Framework?
CRIT stands for four steps: Context, Role, Interview, Task. Woods calls it the backbone of 99 percent of his AI prompts. It’s not complicated. It’s not technical. It’s a repeatable process that forces you to give AI enough information to actually help you.
Here’s the one-sentence version of each step.
Context means giving AI your world. Your business, your industry, your goals, your audience. One paragraph that sets the table.
Role means telling AI who to be. Not just “be helpful,” but a specific expert. A CFO analyzing your numbers. A brand strategist evaluating your positioning. A marketing director planning your next quarter.
Interview means letting AI ask you questions before it gives you answers. This is the step most people skip, and it’s the most powerful one.
Task means defining exactly what you want. Not “help me with marketing” but “give me five email subject lines for a product launch targeting restaurant owners in the Southeast.”
That’s it. Four steps. The magic isn’t in any single step. It’s in using all four together so AI has the context, perspective, and clarity it needs to give you something worth reading.
How Context Changes Everything
Context is the foundation. Without it, AI fills in the blanks with assumptions, and those assumptions are always generic.
When you type “write me an email to my customers,” AI doesn’t know if you sell software or sandwiches. It doesn’t know if your customers are Fortune 500 executives or first-time homeowners. It doesn’t know if you’re announcing a price increase or a holiday sale. So it writes something that could apply to anyone. Which means it applies to no one.
A context paragraph looks like this: “I own a 12-person marketing agency that works with small businesses in the restaurant and hospitality industry. Our clients are typically owner-operators doing $500K to $3M in annual revenue. We’ve been in business for seven years and our main differentiator is that we handle branding, web design, and ongoing content marketing under one roof.”
That took 30 seconds to write. But now every response AI gives you is filtered through your actual reality. The advice gets specific. The language matches your industry. The examples make sense for your situation. You went from asking a stranger for directions to giving them your address.
If you’ve ever wondered why your brand messaging feels disconnected from what AI generates, this is almost certainly the reason. AI can’t match your voice or your market if it doesn’t know who you are.
Why Assigning a Role Matters More Than You Think
Most people skip the Role step because it feels unnecessary. You’re already talking to AI. Why tell it who to be?
Because the role changes the lens. When you tell AI to respond as a CFO, it prioritizes numbers, risk, and financial impact. When you tell it to respond as a brand strategist, it prioritizes positioning, differentiation, and audience perception. Same question, completely different answers, because the expertise filter changes what gets emphasized.
Try this experiment. Take any business question you’ve been thinking about. Ask AI the question once with no role assigned. Then ask it again with this prefix: “You are a strategic marketing consultant with 20 years of experience working with small businesses. You specialize in helping companies under $5M in revenue build brands that attract premium clients.”
The first answer will be a textbook summary. The second will read like advice from someone who actually understands your world. The words change. The priorities change. The specificity changes. All because you told AI which expert to channel.
Woods recommends getting specific with roles. Not just “be a marketing expert” but “be a direct-response copywriter who specializes in email campaigns for service-based businesses.” The more specific the role, the more targeted the output. It’s the difference between asking “a doctor” versus asking “a pediatric cardiologist.” Both are qualified, but one gives you exactly what you need.
What Makes the Interview Step So Powerful?
The Interview step is where CRIT separates itself from every other prompting framework out there. Most frameworks tell you to give AI information and then ask for output. CRIT adds a middle step: let AI ask you questions first.
Here’s why this matters. You don’t always know what information is relevant. You might think AI has everything it needs after your context paragraph and role assignment. But there are details you take for granted about your business that AI doesn’t have. Your pricing model. Your biggest competitor. The objection you hear most on sales calls. The seasonality of your revenue. These details shape the quality of the output, and you might not think to include them.
When you add “Before you complete this task, ask me up to five questions, one at a time, to gather what you need” to your prompt, something interesting happens. AI starts interviewing you. It asks about your budget. Your timeline. Your past results. Your constraints. And as you answer those questions, two things happen simultaneously: AI gets better information, and you get clearer on your own thinking.
Woods calls this the most powerful and most skipped step in the framework. Business owners rush past it because they want the output. But the interview is where the real value gets created. It turns a transaction into a conversation. And conversations produce better results than transactions every single time.
There’s a bonus effect here that’s easy to miss. The interview step often surfaces problems you didn’t know you had. A business owner asking AI for help with a sales email might get an interview question like “What’s the primary objection your prospects raise before buying?” If you can’t answer that clearly, you’ve just discovered a gap in your sales process that no amount of AI copywriting will fix. The framework doesn’t just improve AI output. It forces you to think more clearly about your own business.
Defining the Task So AI Actually Delivers
The last letter in CRIT is the one that seems obvious but trips people up the most. The Task step isn’t just “tell AI what to do.” It’s about defining the deliverable with enough specificity that you don’t have to rewrite the output from scratch.
A weak task looks like this: “Write me some marketing content.”
A strong task looks like this: “Write five email subject lines for a Black Friday promotion targeting restaurant owners. Each subject line should be under 50 characters, create urgency without using the word ‘hurry,’ and reference the pain of slow weeknight traffic.”
The difference is specificity. You’ve told AI the format (email subject lines), the quantity (five), the audience (restaurant owners), the constraints (under 50 characters, no “hurry”), and the angle (slow weeknight traffic). AI doesn’t have to guess at any of those decisions. You made them in advance.
Good task definitions include four things: the deliverable type (email, blog outline, ad copy, strategy memo), the scope (how many, how long, how detailed), the audience (who will read or use this), and any constraints (word count limits, words to avoid, tone requirements). Think of it like a creative brief. The tighter the brief, the better the work.
If you’ve ever struggled to get your website copy right, this step alone would save you hours. Instead of generating generic page content and rewriting it line by line, you define exactly what you need before AI writes a word.
What Does CRIT Look Like in Practice?
The best way to understand CRIT is to see the same prompt written two ways.
Without CRIT:
“Write a blog post about why small businesses need a website.”
You’ll get a 500-word summary that sounds like it was pulled from a 2019 marketing textbook. It’ll mention mobile responsiveness, credibility, and “your online presence.” Nothing wrong with it. Nothing useful about it either.
With CRIT:
“Context: I own a full-service marketing agency that works with small businesses across restaurants, law firms, healthcare, and professional services. Our typical client is an owner-operator doing $500K to $3M in revenue who knows they need marketing help but feels overwhelmed by options. We build brands, design websites, and create ongoing content strategies.
Role: You are a direct-response content strategist who specializes in writing blog posts that convert readers into leads for service-based businesses. You write in a conversational, no-jargon style that prioritizes specific numbers and real examples over vague claims.
Interview: Before you write, ask me up to 5 questions, one at a time, to gather the context you need to write a high-performing post.
Task: After the interview, write a 1,200-word blog post targeting small business owners who are unsure whether to invest in a professional website. The post should include at least three specific statistics, address the top two objections to spending money on a website, and end with a clear call to action to schedule a consultation.”
The first prompt gets you filler. The second gets you a draft you might actually publish. Same AI tool. Same topic. Completely different results. The only variable is how you structured the ask.
This isn’t theory. Woods reports that this framework is the backbone of his approach across every business application, from strategy sessions to hiring decisions to financial analysis. The CRIT structure works because it mirrors how you’d brief a real expert. You’d tell them about your business, explain what kind of thinking you need, answer their questions, and then tell them what to produce. CRIT just makes that process repeatable.
Where Should You Start?
If you’ve been using AI casually and getting mediocre results, start with one prompt today using the CRIT structure.
Pick a task you’ve been procrastinating on. Writing an email to your customers. Drafting a social media plan. Figuring out your pricing strategy. Something you’d normally spend an hour stalling on before producing something you’re not thrilled with.
Write one paragraph of context about your business. Assign a role that matches the expertise you need. Add one line asking AI to interview you with up to five questions. Then define your task with a specific deliverable, audience, and format.
That’s it. One prompt. One task. And when you see the difference in what AI gives you back, you’ll understand why 58 percent of small businesses are using these tools and why a good chunk of them are still frustrated. The tool works. The framework is what makes it work for you.
If you want to go deeper, pick up Geoff Woods’ book The AI-Driven Leader. It covers the full framework with examples across strategy, operations, hiring, and leadership decisions. For a book about AI, it’s surprisingly practical. No hype, no jargon, just a repeatable system that makes AI useful for the decisions you actually make as a business owner.
The business owners who get the most out of AI aren’t the ones using the fanciest tools. They’re the ones who learned how to ask better questions. CRIT gives you a structure for asking better questions every time, whether you’re drafting marketing emails, planning your next quarter, or trying to figure out why your website isn’t converting the way it should.
Your next AI prompt will either sound like every other generic request, or it’ll sound like a strategic brief from someone who knows exactly what they need. CRIT is the difference. And now you know how it works.