Day 3 of 20 · AI for Recruitment

Writing Job Descriptions That Attract

⏱ 7 min 📊 Beginner

A job description isn't a legal document. It's a marketing campaign.

Every JD you post is competing with dozens of others for the same candidates. The ones that attract the best people are clear, compelling, and inclusive. The ones that repel good candidates are bloated with jargon, unrealistic requirements, and corporate boilerplate.

Today you'll learn how to use AI to write job descriptions that make qualified people actually want to apply — and that don't accidentally filter out great candidates before they even start.

Before and after job description — from corporate jargon to clear, compelling, and inclusive

AI transforms dull job descriptions into compelling ones that attract the right candidates.

Why most job descriptions fail

Let's be honest about the state of most JDs:

They're wish lists, not role descriptions. "Must have 10+ years of React experience" — React has existed since 2013. "Must have experience with [15 different technologies]" — you're describing a unicorn that doesn't exist.

They use coded language that repels diverse candidates. Research shows that words like "rockstar," "ninja," "aggressive," and "dominant" discourage women from applying. Phrases like "young and dynamic team" discourage older candidates. Most of this is unintentional, but the impact is real.

They bury the value proposition. Candidates want to know: What will I work on? Who will I work with? What will I learn? How will I grow? Most JDs don't answer any of these questions until the last paragraph — if at all.

They're recycled from three years ago. The role has evolved, the team has changed, but the JD is the same one your predecessor wrote in 2023 with a few tweaks.

AI doesn't just rewrite your JDs. It catches these problems and fixes them.

Knowledge Check

What is the biggest problem with using words like "rockstar" and "ninja" in job descriptions?

Research shows they discourage diverse candidates, particularly women, from applying

They're outdated and unprofessional

They're too casual for corporate roles

Candidates don't know what they mean

Studies from the Journal of Personality and Social Psychology show that masculine-coded words like "rockstar," "ninja," "dominant," and "aggressive" significantly reduce the number of women who apply. AI can scan your JDs for these coded terms and suggest inclusive alternatives that attract a broader, more diverse candidate pool.

The AI inclusive language checker

One of the most valuable things AI does for JDs is catch biased language you didn't even know was there.

Try this prompt with any JD you're working on:

"Review this job description for gendered language, age bias, ability bias, and any phrasing that could discourage diverse candidates from applying. Flag each issue and suggest an alternative."

Here's what AI typically catches:

Gendered language — "He will manage..." becomes "You will manage..." "Manpower" becomes "Workforce." "Chairman" becomes "Chair."

Unnecessary requirements — "Must have a degree from a top university" — is that actually required, or is it filtering out self-taught candidates who are equally capable? "Must have 8+ years experience" for a mid-level role — research shows women tend to apply only when they meet 100% of requirements, while men apply at 60%.

Ability assumptions — "Must be able to stand for extended periods" — is that actually required for a desk-based role?

Age signals — "Digital native" suggests you want someone young. "High energy" can be coded language for "not old."

Run every JD through this check before posting. It takes 30 seconds and meaningfully improves your candidate pool.

Requirements vs. nice-to-haves

This is where most hiring managers and recruiters get it wrong — and where AI can be your best ally.

Ask AI to help you separate genuine requirements from preferences:

"I'm writing a JD for a Senior Data Engineer. The hiring manager has given me this list of requirements. Help me separate them into 'must-haves' that are genuinely essential from 'nice-to-haves' that would be bonus skills. Be honest about which ones are actually necessary to do the job from day one."

AI will push back on inflated requirements. It might say: "Kubernetes experience is listed as a requirement, but if the primary work is building data pipelines in Python, this is a nice-to-have. Listing it as a requirement will filter out strong data engineers who haven't worked in containerised environments."

This kind of honest analysis typically reduces your requirements list by 30-40% — and dramatically increases the number of qualified applicants.

Knowledge Check

Why is separating "must-haves" from "nice-to-haves" so important in a job description?

It makes the JD shorter and easier to read

It helps the hiring manager sound more reasonable

It reduces the number of applications you receive

Inflated requirements discourage qualified candidates from applying, especially women and underrepresented groups who tend to self-select out when they don't meet 100% of listed criteria

Research consistently shows that women and underrepresented candidates apply only when they meet nearly all listed requirements, while other candidates apply at 60% match. By clearly separating must-haves from nice-to-haves, you signal that meeting every single criterion isn't mandatory — which significantly broadens and diversifies your applicant pool.

Crafting a compelling EVP

Your Employer Value Proposition (EVP) is the answer to the candidate's real question: "Why should I work here instead of somewhere else?"

Most JDs answer this with generic fluff: "competitive salary, great benefits, dynamic team." That tells the candidate nothing. Every company says that.

Use this AI prompt to build a real EVP:

"I'm writing a JD for [role] at [company]. Here's what makes this role genuinely appealing: [list 3-5 honest selling points]. Write an EVP section that's specific, compelling, and avoids generic corporate language. Make it sound like a real person explaining why they love working here."

A strong EVP addresses:

- Impact — What will this person actually change or build?

- Growth — How will they develop? What will they learn?

- Team — Who will they work with? What's the team like?

- Flexibility — Remote, hybrid, hours, autonomy?

- Compensation — Be transparent if you can. Salary ranges attract 30% more applicants.

🤖

ChatGPT

online

Here's a job description our hiring manager wrote for a Full Stack Developer. It's pretty generic and I think it's putting people off. Can you transform it into something that would actually make a good developer want to apply? Keep the core requirements but make it compelling and inclusive. [Pastes generic JD with: "We are looking for a rockstar Full Stack Developer to join our fast-paced, dynamic team. Must have 7+ years experience, CS degree required, expert in React, Node, Python, AWS, Docker, Kubernetes, PostgreSQL, MongoDB, Redis, GraphQL..."]

Here's your transformed job description: **Full Stack Developer — Build the Platform 2M Users Rely On** You'll join a team of 6 engineers building and scaling our customer platform. This isn't maintenance work — we're rebuilding our core product, and you'll have real ownership over architecture decisions. **What you'll do:** - Design and build features across our React frontend and Node.js backend - Work directly with product and design — no tickets thrown over a wall - Make technical decisions that affect how 2 million users experience our product - Mentor junior developers and shape our engineering culture **What you need:** - 4+ years building full stack applications (React + Node.js preferred) - Comfort with relational databases (we use PostgreSQL) - Experience deploying to cloud environments (AWS preferred, others fine) - Clear communication — you'll be explaining technical trade-offs to non-technical stakeholders **Bonus if you have:** - Experience with containerisation (Docker, Kubernetes) - Familiarity with GraphQL or message queues - Contributions to open-source projects **What we changed:** Removed "rockstar" (gendered language), reduced the requirements from 10 technologies to 4 actual needs, removed the CS degree requirement (which filters out talented self-taught developers), lowered years from 7+ to 4+ (which is realistic for the actual work), and led with impact instead of demands. The original would attract a narrow pool. This one invites strong developers who can actually do the job.

↻ Replay conversation

AI prompt templates by role type

Here are ready-to-use prompts for different kinds of roles:

Technical roles: "Write a job description for a [role] at [company]. Focus on the technical challenges they'll solve, not just the tech stack. Include what they'll build in their first 6 months. Separate must-have from nice-to-have skills. Use inclusive language."

Sales roles: "Write a job description for a [role] at [company]. Lead with the earning potential and career path. Describe the product/market and why it's sellable. Include realistic OTE and what top performers earn. Avoid aggressive language."

Executive positions: "Write a job description for a [role] at [company]. Focus on the strategic challenges and what success looks like in 12 months. Emphasise the scope of impact and the team they'll build. Keep requirements focused on leadership competencies rather than specific technical skills."

Graduate schemes: "Write a job description for a [role] graduate scheme at [company]. Emphasise learning, mentorship, and career progression. Be specific about what the programme includes. Avoid requiring experience — focus on potential, curiosity, and transferable skills."

Save these templates. You'll use them constantly.

Knowledge Check

When writing a JD for a graduate scheme, what should you emphasise?

Specific technical certifications

Years of experience required

The prestige of the company brand

Learning opportunities, mentorship, career progression, and potential over experience

Graduate candidates are looking for growth, learning, and a clear career path. They don't have years of experience to list. Emphasising mentorship, programme structure, and what they'll learn — while focusing on potential and transferable skills — attracts a much stronger and more diverse pool of early-career talent.

The before-and-after test

Here's a simple habit that will improve every JD you write:

Before you post a job description, ask AI to do a final review:

"Read this job description as if you're a qualified candidate considering whether to apply. What would excite you? What would put you off? What's unclear? What's missing? Be brutally honest."

This 30-second check catches issues your eyes have glazed over. AI might flag that you forgot to mention remote work options, that the "About Us" section is longer than the role description, or that a requirement feels unrealistic for the level you're hiring.

Then ask: "Now read it as an underrepresented candidate who's slightly unsure if they're qualified enough. What would make them decide not to apply?"

That second question is gold. It surfaces the subtle signals that limit your pipeline without you realising it.

👥

Day 3 Complete

"A job description is a marketing campaign, not a requirements dump. Use AI to write inclusive, compelling JDs that attract the candidates you actually want — then check them for bias before posting."

Tomorrow — Day 4

Boolean Search & AI Sourcing

Tomorrow you'll use AI to generate powerful boolean search strings and source candidates across LinkedIn, GitHub, and beyond.

Continue to Day 4 →