The Niche Job Board: A Small Business You Can Start With a Spreadsheet

A friend of mine spent three weeks last spring trying to hire one person. Not a unicorn engineer — a part-time bookkeeper who understood veterinary clinics. She posted on the big job sites, got 180 applications, and roughly four of them were from humans who had read the description. The rest were spray-and-pray résumés fired off by people applying to everything with a pulse.

That gap — between "a lot of candidates" and "the right candidates" — is a business. It has been a business for twenty years, and it keeps being one because the giant job platforms are optimized for volume, not fit. The niche job board is one of the last small-business ideas where a single person, working evenings, can build something that pays real money and keeps paying while they sleep.

Here's the idea in one line:

Pick a job market so specific that both sides feel underserved, put the two sides in one room, and charge the side with the budget.

Why a tiny job board still works in 2026

The intuition most people have is that this space is finished. Indeed, LinkedIn, and the aggregators own hiring; what's left for a person with a laptop? The answer is that scale is exactly what makes them bad at niches. A general platform must serve every industry, so its filters are generic, its candidate pool is diluted, and its ranking is tuned for the median employer. If you hire vet-clinic bookkeepers, "median" is useless to you.

Employers feel this as a cost problem, and that's the crack you slip through. Industry write-ups this year put per-post pricing on niche boards somewhere between $100 for broad boards and $300–$600 for specialized technical niches, with executive-level boards going past $500 — and the reason employers pay it is arithmetic, not generosity. If a bad hire costs tens of thousands of dollars and a wasted month, a few hundred dollars for twenty pre-filtered applicants is cheap. Pricing tends to track two things: the average salary of the roles you list and how scarce qualified candidates are. High salaries plus scarcity equals pricing power. That's your whole selection criterion for a niche.

The scale you need is also smaller than people assume. Public numbers floating around this year include an AI/ML-focused board doing roughly $2.3K/month at near-total margin, and RemoteOK — a one-person-origin project — reported around $35K/month. Neither of those requires a team. They require a specific audience and a few years of consistency.

Choosing a niche that can actually pay

Most failed job boards die at this step. Someone picks "remote jobs" or "startup jobs," which is not a niche — it's a category with a dozen well-funded incumbents. A workable niche passes four tests.

It has money on the employer side. Hiring must be painful and expensive. Boards for hourly retail work can charge around $50 a post; boards for specialized engineers or licensed professionals charge ten times that for the same amount of work on your end. Same effort, different zero.

It has a place where the candidates already gather. A subreddit, a Discord, a professional association, a conference, a newsletter. If you cannot name where your future candidates hang out today, you have no distribution and you will be shouting into a void.

It's specific enough to be describable in five words. "Jobs for veterinary practice managers." "Compliance roles at fintech startups." "Remote roles for medical writers." If you need a paragraph to explain who it's for, employers won't know whether to buy.

You have some unfair proximity to it. You worked in it, married into it, volunteer in it. Insider vocabulary is what makes a listing page feel native instead of scraped.

The lean build: two weekends, not two quarters

This is the part where would-be founders disappear into a six-month coding project and never come back. Don't. The first version of a job board is a list, and a list can be a newsletter.

The staircase looks like this:

Stage 1 costs nothing but attention. You scrape and curate by hand — yes, by hand — because the manual version teaches you the market's vocabulary, salary bands, and which employers are chronically desperate. Those desperate employers are your first customers.

If you do want a little code early, a scheduled fetch that pulls candidate listings for you to review is enough:

import requests, datetime

FEEDS = [
    "https://example-company.com/careers.json",
    "https://another-employer.com/jobs/feed",
]
KEYWORDS = ("veterinary", "practice manager", "clinic")

def looks_relevant(job):
    text = f"{job.get('title','')} {job.get('description','')}".lower()
    return any(k in text for k in KEYWORDS)

picks = []
for url in FEEDS:
    try:
        for job in requests.get(url, timeout=10).json():
            if looks_relevant(job):
                picks.append(job)
    except Exception as e:
        print(f"skip {url}: {e}")

print(f"{datetime.date.today()}: {len(picks)} candidates to review by hand")

That's it. A human still decides what goes in the issue. Curation is the product.

How the money actually arrives

There are more revenue lines than the obvious one, and the healthy boards stack them in this order.

Paid listings are the base. Charge per post, 30-day expiry. Start lower than you think — you're selling into an empty room — and raise the price once you can tell employers "the last five posts averaged N qualified applicants." Proof buys pricing.

Featured placement is the easiest upsell in the world: the same listing, pinned to the top, in bold, for a multiple of the base price. It costs you one database column.

Subscriptions come later, when a handful of employers hire continuously. A monthly plan for unlimited or bundled postings turns lumpy income into a floor — even a few subscribers at a few hundred dollars a month meaningfully changes the shape of the business.

Sponsorships in the newsletter monetize the candidate side without charging candidates, which you should never do. Job seekers are the inventory that makes employers pay. Keep that side free forever.

The realistic check

Now the part the "start a job board this weekend" posts skip.

Cold start is brutal, and it's asymmetric. Employers won't pay for an empty board; candidates won't subscribe to a board with no jobs. You break the loop by faking one side honestly: you hand-collect real, publicly posted jobs from company career pages, list them for free with a link back to the employer, and grow the candidate list on that. Nobody is deceived, and employers eventually notice inbound applicants saying "I found this on your board."

It's slow. Expect six to twelve months before the first invoice feels routine. The compounding asset is the candidate list, and lists grow linearly at first and only later exponentially. If you need income in ninety days, this is the wrong idea.

It's a grind of small, boring tasks. Reviewing listings, chasing dead links, emailing hiring managers, writing a weekly issue on a Sunday night when you'd rather not. The moat is precisely that most people quit this part — but be honest with yourself that "boring and repetitive" is the job description.

Defensibility is thin early on. Anyone can copy your site in a weekend. What they cannot copy is the 4,000 people who open your email and the twelve employers who trust you. So spend your effort on the list and the relationships, not on the design.

The takeaway

A niche job board is a small, unglamorous, quietly durable business. Pick a market where hiring is expensive and candidates are scarce; start as a hand-curated list, not a platform; keep it free for job seekers and charge the employers who are already burning money on bad applicants; then stack featured posts and subscriptions on top of basic listings. Prices and platform numbers cited here are as of writing and vary widely by industry — treat them as a map, not a guarantee.

The best part is that the failure mode is cheap. Worst case, you spend a few months building an email list in an industry you know well and learn who's hiring in it. That's not a bad consolation prize for a business that costs almost nothing to start.