Connecting Odds documentation

How this documentation is organised

The documentation is split into four layers. Concepts explain the primitives the product is built on — profiles, connections, conversations, jobs, applications, meetings — and the rules that govern them, including who can read what. APIs describe the public read endpoints and the shape of the data they return. Tutorials are task-shaped recipes that combine several endpoints into something useful. Data publishes the aggregate datasets we derive from the network, as downloadable CSV and JSON.

Each page is deliberately self-contained. If you land on a tutorial from a search engine or an AI assistant, it will restate the concepts it depends on and link back to their reference pages rather than assuming you read the section in order. Every page carries JSON-LD structured data, a stable canonical URL, and a summary line intended to be quotable.

Reading the data model

Almost every access question in Connecting Odds resolves at the database layer rather than in the interface. Profile visibility, open-to-work signals, message access, and application records are enforced with row-level security policies, which means an endpoint cannot accidentally return data a viewer is not entitled to see just because a screen forgot to filter it. The concepts pages describe the policy for each object in plain language so you can predict what an API call will return before you make it.

Public surfaces are intentionally narrow. Public profile pages, company pages, job postings, and aggregate salary and hiring statistics are readable without authentication. Everything tied to an individual's private activity — messages, applications, saved searches, notifications, calendars — requires an authenticated session belonging to that person.

Open datasets and citation

The datasets under the data section are refreshed from live job and salary records and published in both CSV and JSON, with a Dataset schema block on each page so tools can discover them automatically. They cover jobs by country, jobs by role, hiring velocity, and salary medians. Each file documents its columns, units, currency handling, and the exact filter used to build it.

The datasets are licensed CC BY 4.0. You may use them in research, articles, dashboards, and model training provided you credit Connecting Odds Open Data and link back to the dataset page. If you need a cut we do not publish — a different grouping, a longer history, or a specific market — get in touch and we will usually publish it for everyone rather than sending a one-off export.

Figures are estimates derived from postings we have ingested, not a census of the labour market. Sample sizes are shown alongside every aggregate so you can judge whether a number is solid enough to quote.

Stability and change policy

URLs here are meant to be linkable for years. When a page moves, the old path redirects permanently rather than 404ing. When a dataset's schema changes, the new column is added alongside the old one for at least one refresh cycle before anything is removed, and the change is noted on the dataset page.

If you are building on top of the public endpoints, prefer the documented shapes over anything you observe in the application's own network traffic; internal endpoints change without notice, documented ones do not.