A good prompt is clear, specific, and easy to answer with a simple “yes” or “no.”
A great prompt will:
Say exactly what you’re looking for
Specify the context (e.g. professional vs academic)
Make it something that can clearly be answered yes or no
Avoid:
Single words with no context
Vague terms like “strong”, “good”, “solid”
Multiple questions in one prompt (e.g. “Has management experience and knows React”)
Tip: put yourself in the shoes of the AI. If you were given the prompt and the context (CV, Q&A's, JD, etc), how would YOU interpret the task? What ambiguity would you be confused by? How could you misinterpret the instructions?
Examples
Let's take a look at some examples around AI fields, evaluating a candidate's experience with JavaScript.
Bad prompt
JavaScript
What does this mean?
Any mention of JavaScript on the CV?
Backend or frontend?
Professional experience or a weekend project?
Does one tiny personal project count?
Better prompt
The candidate has professional experience using JavaScript.
This is clearer, but still vague about what “professional” means.
Great prompt
The candidate has professional experience using JavaScript. This means they have used it in a paid role. For example, a university project would not count, but an internship where they used JavaScript would count.
This:
Defines what counts (paid work, including internships)
Defines what doesn’t (university projects / personal projects)
Can be answered yes/no for each candidate
Other good examples:
The candidate has experience managing people. This means they have had direct reports in a paid role (e.g. line management of at least one person), not just mentoring or coaching informally.
The candidate has experience selling B2B SaaS products. This means they have sold software-as-a-service to other businesses as part of their core role, not just used SaaS tools.
The candidate has experience working in a startup environment. This means they have worked for a company with fewer than 100 employees, not just collaborated with startups as clients.
The candidate has experience writing production SQL queries. This means they have used SQL regularly in a job to query live or reporting databases, not just completed a single course or one-off project.