The trade-off behind every entry
There is a real choice in any directory project: scale fast and accept that some entries will be wrong, or scale slow and refuse to ship anything we cannot defend. Most aggregator sites take the first path because the unit economics push them there — every URL is an SEO surface, and the marginal cost of a fabricated email is zero. We chose the second path, and the structural choice runs through every part of the codebase.
The cost shows up immediately. When we ship the Jobs by Country registry, we do not ship 19,700 entries on day one. We ship five countries deep, then expand at a sustainable rate of five to ten verified countries per release. This is slower than the "AI-generated full coverage" alternative; it is also the only honest way to make the standing claim that every URL on the site points to a real organisation.
What we mean by "real"
A URL on this site has cleared three checks before it appears in a registry:
- It resolves. The link opens to a real, currently-online page on the organisation's canonical domain. Dead links and parked domains are scrubbed.
- The organisation is real. Not a private label, not a content farm, not a domain-squatter. It has a verifiable institutional identity — a regulator, a registration, a peer body, or long-standing community recognition.
- The relationship is what we say it is. If we describe a URL as "official portal", it is the organisation's own canonical site. If we describe it as "professional body", it is the recognised peer association in its field. We do not relabel an aggregator as an authority.
The contact-pattern policy
Every email on this site falls into one of three categories, clearly labelled wherever they appear:
Verified. The address has been looked up against the live organisation's site or contact page within the last 30 days. Examples: [email protected], [email protected]. These are kept under freshness review and re-verified every quarter.
Typical. The address follows standard organisational contact convention (info@, careers@, contact@, support@) and is highly likely to be correct, but has not been independently confirmed during the last review window. We mark these explicitly as typical pattern rather than verified.
Not published. When no public organisational contact could be confirmed, the contact field is blank. We do not fabricate or guess. A blank field is a feature, not a missing data point.
The result is a directory where, when you cold-email a contact, you have a clear expectation of what you're using: a freshly-verified address, a standard pattern likely to land, or no email at all.
Why we use handwritten guidance not template prose
Many directory sites pad each entry with template-generated prose — "Welcome to {{country}}, where {{adjective}} opportunities await {{audience}}." The result reads passably but contains no information that is not also in the entry's structured fields. We do not ship that. Every guidance paragraph in the project — the country tldrs in Students by Country, the topic tldrs in Nomad Oasis, the introductory commentary on each subject resources page — is written by a human (one of the principals or with their direct review) and contains specific, opinionated, often falsifiable claims.
This is slower to scale, and it is also the reason readers come back. A directory of links is replaceable; a directory of links plus authentic interpretation is not.
How registries grow
Every shipped version of the site must demonstrate three things:
- URL count increases. No version may regress. New entries are additive — old entries are corrected or marked stale, not deleted.
- Data-point count increases. Every entry's structured fields are filled in deeper over time. A country page that had 5 sections last quarter has 7 this quarter, etc.
- Quality bar rises. The verification standard for new entries must equal or exceed what was applied to the prior batch. If the bar is rising, the rate of growth slows; that's expected and encouraged.
What we do not do
We do not run any external API calls at request time. Every URL you visit composes its content from local PHP data files in a deterministic way. This means you never wait on a flaky third-party, you never see an outage cascading from someone else's outage, and we never silently change content based on data that arrived after a page was indexed. Background freshness updates run on cron, write to local data files, and only become visible when the next batch ships.
We do not run advertising. We do not run behavioural retargeting. We do not embed third-party analytics beyond a single GA4 property used for understanding aggregate traffic patterns. There are no affiliate links unless explicitly disclosed, and there are no paid placements anywhere in any registry.
How to verify
Pick any 5 random rows from any registry on the site. Cross-check every URL by visiting it in a private browser session. Cross-check every contact address by visiting the organisation's official contact page. If you find a single fabricated entry, write to [email protected] and we will publish a public correction.
This standard binds us. It is the reason this project takes longer than it needs to, and the reason we believe it will outlast directory projects that took the faster route.