Fix Poor Job Descriptions with Claude-Powered Talent Acquisition

Use Structured Intakes as the Starting Point for Claude

The quality of Claude’s output depends heavily on the inputs it receives. Replace unstructured emails and chat messages from hiring managers with a simple, structured intake template. Capture fields like role purpose, top 5 responsibilities, success metrics after 6–12 months, must-have skills, nice-to-have skills, location/remote setup, and compensation band (if shareable).

Once you have this structure, feed it directly into Claude with a clear instruction. For example:

You are an HR talent acquisition assistant helping to create a clear, inclusive job description.

Here is the structured role intake:
- Role title: [TITLE]
- Team / department: [TEAM]
- Role purpose (2–3 sentences): [PURPOSE]
- Top 5 responsibilities: [RESPONSIBILITIES]
- Success after 12 months looks like: [SUCCESS]
- Must-have skills/experience: [MUSTHAVES]
- Nice-to-have skills/experience: [NICEHAVES]
- Location / remote policy: [LOCATION]
- Employment type & seniority: [TYPE/SENIORITY]

Tasks:
1. Draft a job description using this structure:
   - About the role
   - Key responsibilities
   - Your profile
   - What we offer
2. Use clear, simple language (B2 English) and avoid internal jargon.
3. Write in a neutral, inclusive tone.
4. Highlight 3–4 concrete impact points to make the role attractive.

This simple intake-plus-prompt pattern can standardise job descriptions across teams while still reflecting the unique aspects of each role.

Build an Internal Job Description Style Guide into Claude

To keep consistency across all postings, turn your existing brand and HR guidelines into a reusable system prompt for Claude. Document your preferred tone, banned phrases, required legal statements, and how you describe benefits and ways of working.

Then embed it in your prompts like this:

You are the internal job description assistant for [Company Name].

Follow our style guide:
- Tone: professional, friendly, direct, no buzzwords.
- Avoid: "rockstar", "ninja", "digital native", age-related phrasing, gender-coded words.
- Always include: diversity & inclusion statement, reasonable accommodation sentence.
- Benefits section must mention: learning budget, flexible work policy, [OTHER KEY BENEFITS].

Given the draft below, rewrite it to match the style guide and flag any potential bias or unclear requirements.

Draft:
[PASTE DRAFT JD HERE]

Over time, you can refine this style guide based on recruiter and hiring manager feedback, effectively training Claude on what "good" looks like in your organisation without needing to build a custom model from scratch.

Add an Automated Bias and Clarity Review Step

Use Claude as a second pair of eyes for every job description before it goes live. The goal is not only to make the text sound nicer, but to systematically catch biased wording, unnecessary barriers, and vague or inflated requirements that can repel strong candidates.

A practical review prompt might look like this:

You are an expert in inclusive hiring and job description design.

Review the job description below and provide:
1. A list of phrases that may be biased or exclusionary, with suggested alternatives.
2. Requirements that could unfairly limit candidates (e.g. unnecessary degrees, years of experience).
3. Sentences that are unclear or full of internal jargon, with clearer rephrasings.
4. A concise "before vs after" example for the 3 most important improvements.

Job description:
[PASTE JD HERE]

Recruiters can then quickly apply or discard Claude’s suggestions, keeping control while making it much easier to maintain high standards of clarity and inclusion at scale.

Create Multiple Variants for Different Channels and Segments

One size does not fit all in job advertising. Use Claude to generate targeted variants of the same role for different talent segments or channels while keeping the core requirements and legal aspects identical. For instance, you might want one version optimised for LinkedIn, another for your career site, and a shorter, impact-focused version for employee referrals.

You can achieve this with a simple variant prompt:

You are helping tailor this job description to different channels.

Base job description:
[PASTE MASTER JD]

Create 3 variants:
1. "Career site" version: full detail, emphasise culture, development and impact.
2. "LinkedIn" version: concise, strong opening hook, focus on key responsibilities and growth.
3. "Referral" version: short, written so an employee can easily forward to a friend.

Keep all requirements consistent with the base job description. Do not add or remove formal criteria.

This approach helps you test which variants perform best without multiplying manual work for recruiters.

Connect Claude to Historical Hiring Data (Even Manually at First)

To move beyond "good sounding" copy and toward better hiring outcomes, give Claude access to your own signals of success: shortlists that hiring managers liked, candidates who passed assessments, or profiles that converted from application to hire. Even without full technical integration, you can start by manually summarising patterns and feeding them into Claude as context.

For example:

You are helping refine a job description based on what has worked in previous hires.

Here are patterns of successful hires for this role type:
- Backgrounds: [e.g. B2B SaaS, mid-market sales, etc.]
- Common skills assessed as strong: [...]
- Red flags identified during interviews: [...]
- Typical career trajectories of top performers: [...]

Here is the current job description draft:
[PASTE JD]

Tasks:
1. Suggest 3–5 adjustments to responsibilities and requirements to better reflect these success patterns.
2. Recommend 2–3 screening questions to include in the application form.

This gradually tunes your job descriptions toward the kind of candidates who actually succeed in your environment, not just those who look good on paper.

Measure Impact and Close the Feedback Loop

Finally, treat Claude-enabled job descriptions as a measurable change in your talent acquisition funnel, not just a cosmetic improvement. Track baseline metrics for a few weeks (time spent per JD, click-to-apply rate, proportion of qualified candidates, hiring manager satisfaction scores), then compare after rolling out the new workflow.

Use Claude itself to help analyse patterns by feeding in anonymised performance data and asking for hypotheses on what to adjust next. For example, you might discover that shorter "Your profile" sections correlate with more diverse applicants, or that more concrete impact statements improve conversion on certain job boards. This continuous improvement loop is what turns an AI text generator into a real recruiting performance lever.

When implemented this way, companies typically see realistic outcomes such as a 50–70% reduction in recruiter time spent per job description, 10–25% improvement in view-to-apply rates on key roles, and a noticeable drop in obviously unqualified applications — freeing talent acquisition teams to invest more time in sourcing, assessment, and offer closing.