Using AI to write a simple position description for a role you've never hired before so you know what you're actually looking for

A bad hire costs roughly 30% of that employee's first-year earnings — for a $50,000 role, that's $15,000 gone before the person has done a single useful thing. This post walks you through a five-step AI workflow that turns your business pain points into a precise position description before you ever write a job posting. Done right, this process takes about two hours and saves you from months of misalignment with a hire who didn't understand the job because you hadn't defined it.

What you need before you start

ChatGPT{:target="_blank"} or Claude{:target="_blank"} — either handles this workflow well. GPT-4o and Claude 3.7 Sonnet (both current as of early 2026) hold context across long conversations and can reformat the same content for different purposes. The free tier of ChatGPT works for this task, but GPT-4o on the Plus plan ($20/month as of early 2026, pricing here{:target="_blank"}) gives you longer context windows, which matters when your brain dump runs long. Claude's free tier is also sufficient for a single session.

Time required: 30–45 minutes for the brain dump and first draft. Add another 30–60 minutes if you go through the challenge round and refinement steps. Budget two hours total for a finished document you'd actually use.

Skill level: No technical setup required. You need to be able to describe your own business clearly in writing — which, if you've been running it, you can do.

---

Why Writing the Position Description First Changes Everything

Most small business owners skip the position description entirely when figuring out how to write a job description for a new role. They copy a generic job posting from Indeed{:target="_blank"} or LinkedIn, change the company name, and call it done. The result: a posting that describes a generic role, attracts generic candidates, and tells you nothing about whether the person in front of you can solve your specific problems.

Here's the catch: a job posting and a position description are not the same document. A job posting is a marketing document — its job is to attract candidates. A position description is an internal document — its job is to define what the role actually does. You cannot write a good job posting without the position description underneath it. The position description comes first, always.

A 2023 LinkedIn Talent Solutions report{:target="_blank"} found that 63% of small business hiring managers cited unclear role expectations as a leading cause of early turnover. That number tracks. If you don't know what the role is, the person you hire won't know either, and both of you will spend the first three months figuring it out at the company's expense.

AI is well-suited to this specific problem because it can reverse-engineer a role definition from your symptoms. You don't need to know what the job is called or what the market expects. You need to describe your pain, and the AI handles the structure.

---

The Brain Dump: How to Tell the AI What You Actually Need

The quality of your position description depends almost entirely on the quality of your brain dump. Vague input produces a plausible-sounding but generic output that could describe any business — not yours. Garbage in, garbage out still applies here, regardless of how sophisticated the model is.

Before you open the chat window, write down answers to these four questions:

1. What problem am I trying to solve by hiring this person?
2. Who does this work now, and how many hours per week does it take?
3. What would "great" look like after 90 days?
4. What's my budget range for this role?

Don't overthink the format. A few bullet points per question is enough. The goal is specificity, not polish.

---

How to Write a Job Description for an Unfamiliar Role: The Starting Prompt

Once you have your notes, open your AI of choice and paste the following prompt, filled in with your specifics:

I run a [type of business] with [number of employees and rough revenue context]. I'm considering hiring someone to handle [problem area]. I currently spend approximately [X hours per week] on this myself, and it's taking time away from [higher-value activity]. Here is what I struggle with most: [list 3–5 specific pain points]. My budget for this role is approximately [$X/year or $/hour]. Based on this, what role am I actually describing, and what should their core responsibilities be? Please draft a role scope with a one-sentence purpose statement, 5–7 core responsibilities, and the skills this person must have versus skills that would be nice to have.

After the AI produces its first draft, read it carefully for two things: whether the responsibilities reflect your actual situation, and whether they're written at the right level of abstraction. Responsibilities should describe outcomes ("manage all inbound lead follow-up within 24 hours") not tasks ("check email"). If the draft is too task-level or too vague, tell the AI exactly that and ask it to revise.

---

How to Use AI to Challenge Your Own Assumptions About the Role

The first draft is not the finish line. The most valuable step in this workflow is asking the AI to interrogate its own output — and yours.

After the draft is on the screen, send this follow-up:

Now play devil's advocate. What assumptions am I making about this role that might not hold up? What questions haven't I answered about scope, decision-making authority, or success metrics? What would make this role fail in the first 90 days?

This step consistently surfaces two problems that plague first-time hires: over-scoping (listing 12 responsibilities that would require two people) and under-scoping (a description so vague the candidate can't understand what winning looks like). The AI will flag both if you ask it to. It will also ask you questions you hadn't considered — who this person reports to, what decisions they can make without you, and whether the budget is realistic for the responsibilities listed.

Answer those questions in the chat, and then ask the AI to produce a revised draft incorporating your answers. This back-and-forth typically takes two or three rounds before the scope is actually tight.

---

What a Finished Position Description Should Include

Before a position description becomes a job posting, it should contain these six elements, per SHRM's guidance on role development{:target="_blank"}:

• Role purpose — one sentence describing why the role exists
• Core responsibilities — 5–7 outcomes-oriented statements, not task lists
• 90-day success metrics — specific, measurable indicators that the hire is working
• Must-have vs. nice-to-have skills — separated explicitly, not lumped together
• Decision-making authority — what this person can decide alone vs. what requires approval
• Reporting structure — who they report to and whether they manage anyone

What you leave out of the position description: compensation range, application instructions, company culture language, and DEI boilerplate. All of that belongs in the job posting, which you write after this document is done.

---

Worked Example: Defining a First Marketing Hire for a Small Service Business

Here's a real scenario. A seven-person landscaping company — $800K in annual revenue, operating across three suburbs — has an owner who spends 10 hours a week on what he calls "marketing." He wants to hire his first marketing person. He has no idea what that means.

The brain dump reveals: He's personally answering Google review responses, posting irregularly on Instagram, sending quote follow-up emails manually, and occasionally updating the website. He hasn't run a single paid ad. He's losing roughly $200K in potential upsell revenue annually because he has no system for following up with existing customers on seasonal services.

The AI's first response: It identifies that he's describing a marketing coordinator role, not a marketing strategist. It separates "execution" (scheduling posts, managing reviews, sending follow-up emails) from "strategy" (campaign planning, channel selection, budget management) and asks which one he's actually missing.

The challenge round surfaces: He's been thinking "marketing" when he actually needs customer retention and lead nurturing. The AI flags that a social media manager won't solve his follow-up problem, and that what he's describing is closer to a marketing coordinator with CRM ownership than a creative hire.

The finished position description defines the role as Marketing and Customer Retention Coordinator, with core responsibilities centered on the quote-to-close follow-up sequence, seasonal upsell campaigns to existing customers, and review management — with social media as a secondary responsibility, not the primary one. The 90-day success metric: a documented follow-up sequence running in the CRM, and a 15% improvement in repeat-customer booking rate.

That's a hire he can evaluate against. The original "marketing person" posting would have attracted a recent grad who knew Instagram and nothing about service business retention.

---

What AI Can't Do Here — and the Two Things You Must Check Yourself

The AI will generate a well-structured, internally consistent position description even if your inputs have gaps. It doesn't know what's missing from your brain dump, and it won't tell you the document is incomplete — it will fill in plausible details instead. That's the honest limitation of this workflow.

Two things you must verify manually before this document becomes a posting:

1. Salary benchmarking. AI has no access to your local labor market. A marketing coordinator role that pays $48,000 in Tulsa is a $62,000 role in Denver. Check the Bureau of Labor Statistics Occupational Outlook Handbook{:target="_blank"}, Glassdoor{:target="_blank"}, and LinkedIn Salary{:target="_blank"} for your specific metro area before you commit to a number. Pricing checked against BLS data as of early 2026 — these benchmarks shift annually, verify before posting.

2. Scope realism. Show the finished position description to one or two people who have actually worked in this function, even informally. A freelancer, a former colleague, or a local business owner who has this role filled. Ask them whether the scope is realistic for one person. AI will not catch scope problems that require industry-specific judgment.

---

What to Do Next

Take the finished position description and run it through one more AI prompt: ask it to convert the internal document into a job posting formatted for Indeed or LinkedIn, keeping compensation language neutral until you've confirmed market rates. That conversion takes about 10 minutes and preserves all the specificity you just built.

For related workflows on using AI in your hiring and operations processes, see how to use AI to run a structured interview process for a small business and using AI to write standard operating procedures before your first hire.

---

FAQ

How do I write a job description for a role I've never hired before? Start with a brain dump of your business problems, not a job title. Describe to an AI what work you're currently doing yourself, how many hours it takes, and what good output would look like. The AI will reverse-engineer a role scope from those inputs. Only after that document is complete should you convert it into a public job posting.

What's the real cost of hiring the wrong person for a new role? The U.S. Bureau of Labor Statistics{:target="_blank"} estimates bad hire costs at 30% of the employee's first-year earnings. On a $50,000 salary, that's $15,000 in wasted recruiting time, onboarding costs, and lost productivity — before you account for the delay in actually solving the problem that triggered the hire.

Can I use the free version of ChatGPT or Claude for this? Yes, for a single session this task fits within the free tier of both tools. The trade-off is context length — if your brain dump is detailed and you go through multiple rounds of revision, a free-tier session may lose earlier context. Claude's free tier handles longer conversations more gracefully than ChatGPT's free tier as of early 2026, but either works for a focused two-hour session.

What's the difference between a position description and a job posting? A position description is an internal document that defines what the role actually does — its purpose, responsibilities, success metrics, and authority level. A job posting is a marketing document designed to attract candidates. The position description comes first and forms the factual backbone of the posting. Most small business owners write only the posting, which is why they end up interviewing candidates for a role they haven't fully defined.

How do I know if I'm over-scoping a new role? After the AI produces a draft, explicitly ask it: "Is this scope realistic for one person, and are any of these responsibilities in conflict with each other?" AI is good at flagging over-scoped drafts — 10+ responsibilities, a mix of strategic and execution-level tasks, or skill requirements that typically belong to two different roles. Then verify the AI's assessment with someone who has worked in the function, since the model won't catch industry-specific judgment calls.