The first marketing job description most founders write is a composite of three real jobs. It asks for a demand generation lead who also writes, owns the website, runs paid acquisition, sets up the CRM, produces the launch and has shipped in your category before. It reads as ambitious. What it actually signals to a good candidate is that nobody has decided what this person is for.
The decision underneath the document is generalist or specialist, and your stage settles it rather than your preference.
Why stage settles it
A specialist is the right hire when you already know which channel is working and need someone to go deeper into it than you can. That is a real situation and it usually arrives after a repeatable acquisition motion exists.
Before that, you do not have a channel problem. You have an unanswered question: which of four plausible channels is worth committing to. Answering it needs someone who can run a small, honest experiment in each, write well enough to make the test fair, and read the result without needing it to come out a particular way. That is a generalist, and a specialist hired into that job will quietly turn it into their speciality because that is what they are good at.
The test is simple. If you cannot name the channel in one sentence, hire the generalist.
Scope the role to the next twelve months, not to an org chart
The most useful thing you can do before writing a word is list what actually has to happen between now and this time next year. Not functions. Outcomes. Something like: the homepage stops being confusing, two of four channels get a fair test, the onboarding email sequence exists, launches stop slipping because nobody owns them.
Then the job description writes itself, because it is that list with a person attached. This also has the useful side effect of showing you when the list is too long for one person, which is a better thing to learn before the hire than after it.
Write the outcomes in, and be concrete about the first ninety days. Candidates worth hiring read that section hardest.
What to leave out
Every tool you have ever used. A tool list filters for people who have used your stack rather than people who can think. Name at most the two that would genuinely be painful to learn on the job.
Years of experience as a proxy for judgement. Five to seven years in a category filters out the operator who spent three years doing exactly your job at a smaller company. Describe the work instead and let people self-select.
The words rockstar, ninja and unicorn. They read as a warning about how the role will be managed.
A secret salary band. In much of Canada you will be posting a range anyway, and a role with no number attached loses the candidates who have options. Check your own province's pay transparency rules before you post, because they differ and have been changing.
Pipeline or revenue targets you have not modelled. A number invented for a job posting becomes a number the hire is measured against in month three.
The honest case for a fractional engagement first
Sometimes the right answer is not to post the role yet. Three situations where that is true:
You cannot describe the job. If the outcome list above came out vague, a hire will inherit that vagueness at full salary and full notice period.
The work is front-loaded. Positioning, a site that explains the product, the first version of the email programme and one launch are a heavy few months followed by a much lighter ongoing load. That shape suits a fixed engagement better than a permanent seat.
You need to know what good looks like before you interview for it. Founders who have watched the work happen once interview very differently afterwards.
A fractional or agency engagement costs more per hour and less per year, carries no notice period of the same weight, and buys you the thing a job posting cannot: a written version of what this role should own. If that is where you land, the retainer that includes fractional product marketing is the shape of engagement I would point at, and I would expect it to make the job description you write afterwards much shorter and much better.
Worth reading alongside this if the decision is still open: the agency, freelancer or first hire comparison, which covers the same three routes for a business of any kind rather than a startup specifically.
I have deliberately kept salary and total-cost numbers out of this piece, because they deserve their own sourcing. The full cost of a hire against a fractional engagement carries the wage data and the employer payroll costs with their verification dates attached. Take the numbers from there rather than from here.
The version worth posting
One paragraph on what the company does, in the words a customer would use. One paragraph on why this role exists now. The outcome list. The first ninety days. Two tools at most. A salary range. A sentence about how the role is likely to change as the company does, because a good generalist wants to know that.
If that document is hard to write, you have found something more useful than a hiring delay. You have found the part of the plan that is not decided yet. What would your outcome list actually say?