Prompt Engineering

A financial analyst at your firm is using Claude to evaluate whether loan applications meet a set of complex, interdependent eligibility rules. The model frequently returns a correct-sounding final verdict but occasionally reaches the wrong conclusion, and the analyst cannot tell where the reasoning went astray. Which prompt engineering technique should the analyst apply first to improve accuracy on this multi-step reasoning task?

A financial analyst wants Claude to evaluate whether loan applications meet a set of five specific eligibility rules. Early tests show Claude sometimes skips rules or gives a verdict without checking each one. The analyst wants to improve reasoning quality but keeps getting inconsistent results when simply saying 'think about it carefully.' Which chain-of-thought refinement is most likely to improve accuracy?

A financial analyst uses Claude to evaluate loan applications. She wants Claude to reason step-by-step through each application's risk factors before giving a final approval decision, but her downstream system only needs the final decision (APPROVE or DENY) so it can be parsed automatically. What is the best prompting approach to satisfy both needs?

A support engineer is writing a prompt for Claude to draft customer email replies. Their current prompt says: "Do not be too formal. Do not use jargon. Do not write more than three paragraphs." The replies are inconsistent — sometimes overly stiff, sometimes cluttered with technical terms. The engineer wants to make the instructions clearer and more reliable without adding examples yet. Applying the 'be clear and direct' technique, what is the single best change to the prompt?

A support engineer is building a prompt that asks Claude to process a customer refund request. The task involves verifying the order ID, checking the return window, calculating the refund amount, and drafting a customer email. In testing, Claude frequently drafts the email before performing the verification and calculation, and sometimes skips the return-window check entirely. The engineer wants to keep this as a single prompt. What is the most effective way to fix this behavior?

A product team asks Claude to "write documentation for our new API." The output is generic and pitched at a level that confuses their audience of non-technical business analysts. The team wants Claude to consistently produce docs the analysts can follow, without switching models or adding examples. Which single adjustment to the prompt best applies the clear-and-direct technique?

A support team asks Claude to "make this customer email better" and pastes a draft reply. The results are inconsistent — sometimes Claude shortens the email, sometimes it adds legal disclaimers, and sometimes it changes the tone entirely. The team wants more predictable output without adding examples or changing models. Applying the 'be clear and direct' technique, what is the best first change to the prompt?

A support team has finalized a prompt for summarizing customer tickets. They want to reuse the exact same prompt structure across hundreds of tickets, changing only the ticket text each time, and manage it centrally in the Anthropic Console. Which Console feature best supports this need?

A team is about to start improving the prompt for a customer-support classification feature. Before they iterate on wording, examples, and XML structure, a senior engineer insists they complete one foundational step first. According to Anthropic's prompt engineering guidance, what should the team do before beginning prompt engineering?

A data analyst is building a workflow where Claude must (1) extract key financial figures from lengthy quarterly reports, (2) analyze those figures for year-over-year trends, and (3) produce an executive summary. When the analyst tries to do all three in a single prompt, Claude frequently drops figures, mixes up the analysis, and produces inconsistent summaries. Which prompt engineering technique should the analyst apply to improve reliability?

A legal-tech team is using Claude to review commercial contracts and flag risky clauses. The output is technically correct but reads like generic advice, missing the precise, cautious tone and terminology a corporate attorney would use. The team wants to shift Claude's responses toward that of an experienced contracts lawyer without adding lengthy new instructions. Which prompt engineering technique most directly addresses this need?

A data analyst is building a prompt that asks Claude to solve multi-variable word problems and return only the final numeric answer. In testing, Claude frequently returns wrong numbers even though the problems are solvable. The analyst wants to improve accuracy without switching models or adding examples. Which single change is most likely to help?

A financial analyst is building a prompt that asks Claude to evaluate whether a loan application should be approved based on several criteria. The analyst wants Claude to reason through each criterion before committing to a final decision, so they add 'Think step by step' to the prompt. However, in testing, Claude often states its approval decision in the first sentence and then writes reasoning that merely justifies that snap decision. What is the most effective adjustment to genuinely improve the quality of Claude's reasoning?

A support team uses Claude to categorize incoming tickets into one of five priority levels. Their prompt includes three examples, all showing clear-cut 'high' and 'low' priority tickets. In testing, Claude handles obvious tickets well but frequently mislabels ambiguous, borderline tickets that fall into the middle priority levels. The team wants to improve accuracy on these harder cases without switching models. What is the most effective adjustment to their multishot examples?

A support team wants Claude to generate replies to customer emails in a very specific style: a warm greeting, a two-sentence apology, and a bulleted list of next steps. Their current prompt describes this style in prose, but Claude's outputs vary widely in structure and tone from one email to the next. The team does not want to change models and has confirmed the task is well-suited to Claude. What is the most effective prompt engineering technique to make the output format and tone consistent?

A support team uses Claude to classify incoming tickets into categories like 'Billing', 'Technical', and 'Account'. With a simple instruction-only prompt, Claude often invents new category names or formats its answers inconsistently. The team wants to improve accuracy and enforce a consistent output style without switching models. What prompt engineering technique should they apply first?

A developer is building a support-ticket classifier. They want to include three labeled examples in their prompt to guide Claude's output. They already have a system prompt defining the classifier role and a user message containing the ticket to classify. Where should the examples be placed to most reliably improve consistency, following Anthropic's multishot prompting guidance?

A developer is building a prompt that must extract meeting action items and format each one as 'Owner | Task | Due Date'. They wrote a long paragraph of instructions describing the exact spacing, capitalization, and how to handle missing due dates, but Claude still produces inconsistent formatting across runs. Which prompt engineering change is most likely to resolve the formatting inconsistency with the least effort?

A developer wants Claude to return only a JSON object with no preamble like "Here is the JSON you requested:". The output is being parsed programmatically and any extra text breaks the parser. Which technique most directly forces Claude to begin its response with the JSON structure itself?

A data analyst is building a prompt that asks Claude to categorize customer support tickets into one of five predefined categories. The analyst wants Claude to reason through ambiguous tickets step by step before committing to a final category, and they want the reasoning captured in a dedicated section so it can be reviewed later but easily stripped from the output. Which prompt engineering technique most directly enables Claude to show its reasoning in a clearly separated, parseable section?