How to Write a Job Posting With AI That Gets the Right Applicants (Not Just More)

Why most small business job postings fail

Chris runs a 6-person marketing agency in Seattle. His job postings consistently attracted 60 to 80 applicants. He consistently hired none of them, extended the search, and eventually promoted from within or asked a contractor to go full time.

The problem was not the candidate pool. It was the posting. Written to sound professional, it was also vague enough that anyone who had ever touched a social media account felt qualified to apply. The posting did not describe what the person would actually do. It described what the company wanted them to be.

AI does not fix a hiring strategy. But it is extremely good at one specific thing: turning a vague, aspirational job description into a specific, honest one that makes the right candidates say "this is me" and the wrong candidates say "this is not for me."

The difference between a filtering posting and a broadcast posting

A broadcast posting: "We are looking for a motivated, detail-oriented marketing specialist with excellent communication skills and a passion for growth."

A filtering posting: "You will own our LinkedIn and email newsletter week-to-week. Monday is content planning. Tuesday and Thursday are writing days. You will pull performance data from HubSpot every Friday and present a one-slide summary at our Monday standup. You are right for this role if you have done this before and liked the repetition of it."

The second posting attracts fewer applications. The applications it attracts are far more likely to be the right people.

The prompt that generates filtering job postings

Answer these questions about your role, then paste them into the prompt below:

• What does this person do in their first two hours on Monday morning?
• What does a bad week in this role look like? What problems do they face?
• What tools will they use daily?
• What would you see in their first 90 days that would make you say "great hire"?
• What type of person has failed in this role or similar roles before?

What to do with the output

Read through the AI draft with one question in mind: "Would a wrong candidate read this and think they are qualified?" If yes, make it more specific. The requirements section especially: if it says "experience with email marketing," change it to "has built and sent email campaigns to a list over 500 people using [specific tool or tools]."

The "What we are like to work for" section will often be generic in the first draft. Rewrite it yourself in one to three honest sentences. This section is where candidates decide whether your culture fits theirs, and it is worth being direct.

The three sections AI writes better than most people

The screening question trick

Add one specific, non-obvious question to the application: "Tell me about a time you found an error in something a manager or colleague had submitted as final. What did you do?" Or: "What is something in this job posting that would frustrate you, and why?"

Candidates who answer this thoughtfully stand out immediately. Candidates who paste a generic response or skip it tell you something important. AI can help you write the question. The answers are all yours to evaluate.

For small businesses that want to track applicants and hiring stages without buying an ATS, HubSpot's free CRM works as a lightweight applicant tracker using its pipeline feature. Notion is another option for teams that prefer a flexible database approach. Current deals on each deal page.