I need to write a job description for a role I'm hiring. Where do I start?

This page may contain affiliate links from which Blur Studio earns commissions.

A job description has two jobs itself: attracting the right candidate, and setting clear expectations once they're hired. A generic AI-written or template description does the first reasonably well and the second badly, because it describes the job title in the abstract, not how your specific business actually splits responsibilities. Both jobs matter, but they pull in slightly different directions, and recognizing that upfront saves you from writing one draft and assuming it's finished, since the draft that attracts the most applicants isn't automatically the one that sets the new hire up to succeed once they actually start the job.

Fast starting points

• ChatGPT (free to low-cost): give it the role title and two or three sentences about your business, and it'll draft a structured description with sections for responsibilities, qualifications and reporting line. Fast, but generic by default until you edit the responsibilities to match your actual workflow.
• A template from SHRM (check current access): professionally structured templates by role category, useful for the format and the standard sections, same caveat on needing real customization of the content.
• Trainual, a platform for writing up and assigning company process documentation, has AI-assisted documentation: drafts role content from a short prompt the same way it drafts process documentation, included unlimited on every plan. The advantage over a one-off AI chat is that the resulting role description lives inside the same platform as your actual process write-ups, so "what this role is responsible for" and "the actual SOPs for doing it" sit next to each other rather than in separate tools. Trainual itself is quote-only across Core, Pro, Premium and Enterprise, with a $1,000 one-time setup fee on new accounts. Its library of role templates is described in Trainual templates.

The edit that actually matters

Whatever draft you start from, the generic responsibilities section is the part that needs real rewriting. "Manages customer relationships" means nothing specific. "Follows up on every quote within 24 hours and logs the outcome in the CRM" tells the actual new hire, and the team, what the job really involves day to day. AI-generated text defaults to the vague version because it doesn't know your real workflow; you have to add that part yourself.

When to skip the AI draft entirely

If the role is highly specific to your business, especially a senior or founder-adjacent position, an AI draft will be more generic than useful as a starting point, and you'll spend more time correcting it than you would writing from your own notes. Use AI for routine, well-understood roles; write unusual or senior ones yourself from scratch.

A worked example: rewriting a generic draft for an operations coordinator

A small logistics business asked ChatGPT for a job description for an "operations coordinator" and got back a competent-looking draft: "coordinates daily operations," "liaises with vendors and internal teams," "ensures smooth workflow execution." Every phrase was technically accurate and told a candidate almost nothing real about the job. The owner rewrote just the responsibilities section by spending twenty minutes listing what the previous person in that role actually did in a typical week: confirms the next day's delivery schedule with three specific carriers by 4pm, flags any shipment running more than two hours late to the sales team directly, reconciles the weekly carrier invoice against the shipment log every Friday. Those three concrete lines replaced "ensures smooth workflow execution" and did two things the generic version couldn't: they let a candidate honestly judge whether they'd actually be good at this specific job, and they gave the eventual new hire a real starting checklist instead of a vague mandate to figure out independently.

The qualifications, reporting line and general formatting from the original AI draft stayed untouched, since there was nothing wrong with them, confirming the useful split isn't "rewrite everything" or "accept everything," it's identifying the one section (responsibilities) that actually needs business-specific detail and leaving the rest as the AI produced it.

A practical note on keeping descriptions and training in sync

A job description and the actual training material for that role tend to drift apart over time if nobody's checking both together: a role picks up new responsibilities that get added to training but never make it back into the official job description, or the reverse, where a posted description lists a duty the role stopped doing months ago. Worth a quick check whenever you're hiring again for a role you've filled before: does the existing job description still match what the SOPs for that role actually say, and if not, which one is the stale one. One way to keep them linked is to assign training by job role.

A practical note on reporting line and growth path

Candidates for mid-level and senior roles specifically weigh who they'd report to and whether there's a visible path beyond the role, more than most owners expect when drafting a first version. A line as simple as "reports directly to the owner" or "reports to the operations manager, with a path to a senior role as the team grows" costs nothing to add and genuinely affects which candidates take the posting seriously, especially for a role you're hoping someone will grow into rather than just fill short term. The same applies to compensation range: a posting that's vague on pay filters out fewer unqualified applicants than a specific range does, and a specific range tends to attract candidates whose expectations already roughly match what you're prepared to offer, saving everyone time later in the process. Setting roles clearly across the company is covered in defining roles and responsibilities.

Common mistakes with job descriptions

• Copying a competitor's job posting almost word for word. It tells you what title and general duties the market expects, but their actual day-to-day split of work is invisible to you, and copying their vague phrasing just imports someone else's generic problem into your own posting.
• Listing every possible task the role could ever touch. A description trying to cover every edge case becomes a wall of text candidates skim past. List the core five or six responsibilities that make up most of a normal week, and address exceptions during the interview, not in the posting.
• Never updating the description once the role evolves. A role's actual responsibilities drift over months as the business changes. Revisit the description whenever you're hiring again for it, rather than reusing a version that's quietly stopped matching reality.

How I'd approach this

I'd draft the title, qualifications and reporting line with AI, since those parts genuinely are fairly standard across businesses, then write the responsibilities section myself by listing the five or six things this specific person will actually do in a normal week. That split gets you speed on the boilerplate and accuracy on the part candidates and new hires actually care about.