Why niche is the only honest answer
A general job board competes with LinkedIn and Indeed, which is not a competition, it is a donation. The boards that work are vertical or local to the point of specificity: "UX jobs in fintech", "brewery work in the Northeast", "remote jobs paying in EUR for LatAm talent". Niche focus is not a limitation on the build; it is the marketing strategy, the moderation strategy, and the pricing power in one decision. Deciding the niche precisely, down to which postings you would reject, is the first deliverable, before any development.
The two-sided shape (smaller than a marketplace, still two-sided)
Job boards are marketplaces with a slower clock: employers post (supply of jobs), seekers apply (demand for them), and the match is asynchronous. The MVP pattern mirrors the marketplace MVP logic: seed the employer side by hand (the first ten paying postings come from outreach, not self-serve), build the seeker side around a single excellent channel (a newsletter is a completely legitimate primary interface), and automate only what volume proves. Application tracking, messaging systems and resume databases are later-version features, the early board can honestly run applications through email without dying of it.
Posting models and the payment wiring
- Pay per post, the workhorse: single postings, 30-day windows, checkout before publish. Simple, honest, easy to sell.
- Packs and subscriptions, five-post packs or monthly plans for recruiters; introduces account state, credit ledgers, expiry logic. Real engineering, justified at real volume.
- Featured/boosted placements, the margin layer on top of either model: pinned, highlighted, newsletter-included. The upsell is where board margin actually lives.
- Free tier, many healthy boards post free internships/community roles as density strategy; the build just needs the moderation to keep "free" from meaning "spam".
All of it rides standard payment plumbing, the Stripe integration pattern covers checkout, webhooks and the state transitions (paid → live → expired) that boards get wrong when wired by hand.
Moderation: the cost that shows up in week three
Public posting invites the internet: spam postings, scam "employers", discriminate-in-the-requirements listings, SEO link farms wearing job descriptions. The moderation load starts light and grows with density, which is why the build includes the tooling (flagging, queue, bulk actions, poster history) and the scoping includes the policy (what gets rejected, how fast, with what appeal). A board's brand is its junk filter; this is a feature line, not an ops afterthought, and quotes that omit it are quoting a different (fantasy) product.
What the build actually contains
- Posting flow, structured job data (not a text blob), preview, payment, publish/expiry windows, edit-after-publish rules
- Seeker side, search and filters per the niche's facets, alerts (email digests are retention gold), apply flow per your policy
- Accounts for both sides, employer dashboards (postings, applicants), light seeker profiles where the model needs them
- Payment layer, per the chosen model, with expiry and renewal mechanics
- Moderation tooling, queue, flags, history, and the policy in writing
- SEO and syndication, structured data for jobs (Google's jobs surface is a real acquisition channel), sitemap hygiene, and optionally backfill from aggregators to fight the cold-start
Honest scope talk: the software is months, the business is quarters, the board lives or dies on the niche decision and the seeding grind. When the build is right-sized to that reality (version one lean, tooling ready), it is one of the better ratio businesses in this whole category. The cost calculator prices a specific configuration; the brief with your niche starts the scoping.