How to Get the Most Out of Claude: Prompting Best Practices for Business Professionals

If you have used both ChatGPT and Claude, you have probably noticed that they feel different. ChatGPT tends to be direct and task-oriented. Claude tends to be more thorough, more nuanced, and more willing to push back if it thinks your framing is off. These are not just personality quirks, they reflect genuinely different design philosophies that affect how you should prompt each tool.

This is Part 2 of our AI Prompting series. Part 1 covered ChatGPT. This guide focuses on Claude by Anthropic, what makes it distinctive, and the specific techniques that consistently produce the best results.

What Makes Claude Different

Claude was designed with a strong emphasis on being helpful, harmless, and honest, and that design philosophy shows up in how it responds. Claude is less likely to produce confident-sounding nonsense. It is more likely to tell you when it is uncertain, to flag assumptions in your question, and to offer a more balanced perspective when you ask for an opinion.

For business use, this means Claude tends to excel at tasks that require nuance: drafting communications where tone matters, analyzing documents with competing considerations, writing content that needs to feel genuinely human rather than AI-generated, and working through complex decisions where the right answer is not obvious.

It also means that some prompting shortcuts that work with ChatGPT, like very terse instructions, work less well with Claude. Claude responds better to context and explanation. The more you tell it about why you need something, the better it performs.

Technique 1: Explain the Why, Not Just the What

Anthropic's own documentation describes Claude as responding well to context and motivation behind instructions. This is not just a stylistic preference, it is a functional difference. When you explain why you need something, Claude calibrates its response to actually serve that goal, rather than just executing the literal instruction.

Instead of: "Rewrite this email to be more professional."

Try: "Rewrite this email to be more professional. The recipient is a CFO who values directness and has limited time. I want to come across as confident and prepared, not apologetic. The goal is to get a meeting on the calendar."

The second prompt gives Claude the context it needs to make good judgment calls about word choice, structure, and emphasis, rather than just swapping casual words for formal ones.

Technique 2: Use XML Tags to Structure Complex Prompts

For complex tasks, analyzing a document, working with multiple inputs, or giving Claude layered instructions, Anthropic recommends using XML-style tags to organize your prompt. This is one of the most distinctive and powerful techniques for Claude specifically.

XML tags help Claude parse your prompt unambiguously, especially when you are mixing instructions, background context, examples, and the actual content you want it to work on. A well-structured prompt might look like this:

<context>
I run a 45-person IT services company in Northwest Indiana. We are preparing a proposal for a healthcare client.
</context>

<document>
[Paste the client's RFP here]
</document>

<instructions>
Identify the top 5 evaluation criteria the client is using, ranked by how much emphasis they place on each. Then suggest how our proposal should be structured to address each criterion directly.
</instructions>

This structure eliminates ambiguity about what is background information versus what is the actual task. Claude handles long, complex prompts well, but only if the structure is clear. XML tags are the most reliable way to provide that clarity.

Technique 3: Give Claude a Role That Matches Its Strengths

Like ChatGPT, Claude responds well to role assignment. But the roles where Claude particularly excels are those that require careful judgment, nuanced communication, or critical thinking. Roles like "experienced editor," "thoughtful advisor," "devil's advocate," or "senior analyst reviewing this for accuracy" tend to produce especially strong results.

Claude is also notably good at playing the role of a skeptic. If you want to stress-test an idea, a plan, or a document, ask Claude to act as a critical reviewer: "You are a skeptical investor reviewing this business case. Identify the three weakest assumptions and explain why each one could undermine the entire argument." This kind of adversarial prompting is one of Claude's most valuable use cases for business professionals.

Technique 4: Use Long Context Effectively

Claude supports very long context windows, meaning you can paste entire documents, lengthy email threads, or multiple files into a single conversation and ask Claude to work across all of them. This is a genuine differentiator for business use cases.

When working with long documents, Anthropic recommends placing the document content at the top of your prompt, above your instructions. This may feel counterintuitive, most people put instructions first, but it consistently improves performance, especially with complex multi-document inputs.

Practical applications: paste a full contract and ask Claude to flag unusual clauses; paste a month of email threads and ask for a summary of unresolved issues; paste a competitor's website and ask for a competitive positioning analysis. These are tasks that would take hours manually and take minutes with Claude when prompted correctly.

Technique 5: Ask Claude to Quote Before It Analyzes

When you are asking Claude to analyze a document, a contract, a report, a proposal, ask it to quote the relevant sections before drawing conclusions. This grounding technique forces Claude to anchor its analysis in the actual text rather than its general knowledge, which produces more accurate and more defensible outputs.

Example: "Review the following service agreement. Before drawing any conclusions, quote the specific clauses that relate to termination rights, liability limits, and data ownership. Then provide your analysis of each."

This technique is particularly valuable for legal, financial, and compliance-related tasks where precision matters and you need to be able to verify the AI's reasoning.

Technique 6: Embrace Claude's Willingness to Push Back

One of Claude's most distinctive behaviors is that it will sometimes push back on your framing, flag a problematic assumption, or offer a perspective you did not ask for. Many users find this frustrating at first. It is actually one of Claude's most valuable features.

When Claude questions your premise, take it seriously. It is not being difficult, it is doing exactly what a good advisor does. If you want to override it, you can: simply acknowledge the concern and restate your request. But often, the pushback reveals a genuine gap in your thinking that is worth addressing before you finalize a document or make a decision.

Technique 7: Iterate with Precision

Claude responds very well to precise refinement instructions. Rather than saying "make this better," specify exactly what dimension you want to improve: "tighten the opening paragraph to one sentence," "replace the third bullet point with a concrete example," "rewrite the conclusion to be more action-oriented."

Because Claude tends to produce longer, more thorough first drafts than ChatGPT, the editing and refinement phase is often where the most value is created. A strong Claude workflow is: generate a thorough first draft, then use targeted refinement prompts to sharpen it into exactly what you need.

A Quick Reference: Claude-Specific Prompt Patterns

The Bottom Line

Claude is not a better or worse tool than ChatGPT, it is a different tool with different strengths. It excels at tasks that require nuance, careful reasoning, and working with long or complex documents. It rewards prompts that include context and explanation. And it is uniquely valuable as a critical thinking partner, a tool that will challenge your assumptions rather than just confirm them.

The techniques in this guide, explaining the why, using XML structure, leveraging long context, asking for quotes before analysis, and embracing pushback, are the habits that separate people who get genuinely useful output from Claude from those who find it frustrating.

This is Part 2 of our 3-part AI Prompting series. Part 3 covers Microsoft Copilot, the AI tool built directly into the Microsoft 365 apps your team already uses every day.

If you want help building a practical AI adoption strategy for your team, including which tools to use for which tasks, contact Schilling IT. We help businesses across Northwest Indiana and Chicagoland use AI tools safely, effectively, and in ways that actually stick.