How the jobs index works: sources, freshness, coverage

Where the adverts come from

There are exactly two routes into the index, and they are not equally good. The large one is a broad sweep of what employers publish openly: 84.2% of all adverts. The small one reads vacancies directly out of employers' applicant systems: 15.7%. Both end up looking identical on the page, and they do not deliver identical information.

The difference runs in both directions, which is why neither route is simply better. The broad route is fresher and far wider: median age 30 days, 23.7% of its adverts a week old or less, and it brings almost five times as many distinct employers into the index as the direct route. It is also the German-language route, with 45.1% of its adverts written in German. The direct route is older but better structured: median age 50 days, yet 98.9% of its adverts carry a summary and 83.7% state a language requirement, against 92.7% and 63.6% on the broad route. It is also the English-language route: 59% of its adverts are in English.

If you want to know where an application actually lands, read the destinations. Every advert in the index carries an apply link, and 45.6% of those links open a named applicant system; 17.7% of all linked adverts land in a system built in Switzerland. By address, 37.3% of the links point at a .ch domain and 43.4% at a .com domain. The single biggest destination across the whole index is a Swiss applicant system, at 7.1%.

What the page shows, and what it never shows

An advert page carries the title, the employer, the location, the employment type, the date and the apply link. Above that, where one exists, sits a section describing the role. That section is our own summary of the advert, and it is present on 93.7% of adverts.

On the remaining adverts the section is simply absent, and that is a decision rather than a gap. The page used to fall back to the employer's own advert text whenever no summary existed; on 1 September 2026 that was happening on more than a thousand published jobs. Reprinting someone else's copy on our domain is a different act from publishing our own summary of it, so the fallback was removed. An advert without a summary now shows the facts and the link and nothing else, until the summary catches up.

The same honesty explains the gaps elsewhere. For 16.2% of published adverts the full text could not be read at collection time. Those rows carry no language requirement and no detected advert language at all — not because the advert asks for nothing, but because we did not read it. That is why every language figure on this page is calculated against the adverts that actually state something, never against the whole index.

How fresh the index is, and what "new" means

An advert's age comes from the date the advert itself states, and 99.7% of adverts carry one. The median is 31 days. 10.9% are three days old or less, 22.8% seven days or less, 46.7% thirty days or less and 79.1% ninety days or less.

Two things you see on the page hang off that date. The "new" flag on an advert card appears up to three days. The count in the page title — "142 new this week" — counts the adverts of the last seven days for that topic. Both are counted from the same rows the page renders, so there is no second definition sitting behind the scenes that could drift away from what you are looking at.

New adverts arrive daily. Because a collection run can also fail quietly, an automatic check runs four times a day and looks at the outcome rather than at the machinery: if the newest advert on the live site is more than 48 hours old, it raises an alarm. That check exists because in August 2026 nothing arrived for nine days and nothing said so.

How an advert leaves the index

A filled role does not disappear on its own. The index re-checks old adverts daily in layers: first whatever the employer's applicant system reports about the vacancy, then unambiguous signs of death on the target page — a 404, a redirect to the homepage, or one of roughly fifty sentences in four languages that say the role is gone. Then comes a deliberate counter-test: if the target page still names the job title, the advert counts as alive, and no later layer is allowed to overrule that.

How well this works was measured on 2 August 2026 against a hundred hand-adjudicated old adverts: 73 of 73 genuinely dead ones were caught, and none of the 13 live ones was wrongly removed. That accuracy has a price. The procedure is deliberately conservative and flags only about a quarter of four-month-old adverts as dead, because a great many closed roles answer with a perfectly healthy careers page that carries no signal left to read. So an age cutoff runs alongside it: an advert nothing has re-confirmed for a long time is treated as closed.

Removal is reversible. The row keeps all of its data and is only hidden, together with the reason it was hidden. Following a link to a removed advert does not land you on an empty page — it takes you to the nearest live list, with a notice explaining the jump instead of leaving it baffling.

What the index covers

Coverage means two different things here: how much of the Swiss market is in the index, and how much of the index has a page of its own. The second question has an exact answer. The catalogue holds 83 roles and 14 cities and regions, crossed across four languages — several thousand topic pages. Not every city appears in every language: the English pages carry 9 of them, the German ones 13. 71.7% of all adverts sit under at least one role page, 65.9% under at least one city page, and 89.7% under at least one of the two. The remaining 10.3% exist in the index but can only be found through search.

Geographically the index leans the way the Swiss labour market leans: Zurich 20.5% of adverts, Bern 8.8%, Geneva 7.2%, Basel 6.2%, Lucerne 4.0% and Lausanne 3.6%. A language could be identified for 76.7% of adverts; among those, 55% are German, 27.1% English, 14.3% French and 2.3% Italian.

Employer concentration is more lopsided than most people expect. 45.7% of the employers in the index have exactly one open advert — and yet the ten largest account for 16.7% of all adverts between them, and the largest hundred for 40.7%. Browsing the index therefore shows you mostly high-volume advertisers, even though nearly half the employers behind it are posting a single role.

Finally, the question job seekers ask most. Excluding apprenticeships, 66.3% of adverts state a language requirement explicitly. Of those, 65% ask for German, 46.3% for English, 22.3% for French and 3.5% for Italian, and 31.3% name more than one language. Apprenticeships are excluded on purpose: they name the qualification they lead to, which distorts any count of requirements.

What the index deliberately does not do

It does not state a salary. Swiss adverts rarely carry one and we do not estimate it. It does not measure workload either: the percentage shown on an advert card is a fixed default, not something extracted from the advert. If a specific workload matters to you, it has to be read in the original advert behind the apply link.

It is not complete, and no private collection can be. One reason is written into Swiss law. For the occupation types on the annually published list, employers must report a vacancy to the regional employment centre before advertising it publicly. That is the entire point of the job-reporting duty: job seekers registered with the RAV are meant to get an information and application head start in occupations with high unemployment. In those occupations a vacancy therefore demonstrably exists at the public employment service before it can exist in any private index, ours included. If your occupation is on that list, look there first.

It does not accept applications. Every apply link leaves for the employer or their applicant system, where their form, their rules and their privacy terms apply. And it does not record job-search efforts at Job-Room on your behalf: filing with the RAV stays your task, and the tracker only fills the form out of the applications you have saved.

What is free, and where the account starts

The whole jobs index works without an account and without paying: every topic page, every page of pagination beneath it, every individual advert page and every apply link. There is no blurred preview here and no signup wall between you and a listing.

The job alert asks for an email address, not an account. It arrives on weekday mornings, carries at most eight adverts, and is only sent when something genuinely new has appeared for your exact search since the last one. The first email looks back 72 hours so that it does not arrive empty.

The account and the paid plan belong to the CV tools, not to the job search. There the ladder is: try anonymously, then a free account with 5 documents a day and 10 per rolling week, then a seven-day trial with full access, then CHF 15 a month or CHF 95 a year, cancellable at any time. Someone who only searches for jobs never meets any of those limits.

How to check all of this yourself

None of this has to be taken on trust. The title of every topic page states how many adverts are open and how many of them arrived in the last seven days — two numbers that have to move between two visits if the index is alive. Every advert card carries its date, and the "new" flag has to agree with that date.

The hard test is the apply link. It goes to the employer's original posting. If the role is still there, the advert is honest; if it is gone, you have found a case our checking has not caught yet — and you now know why that happens. That check is exactly why every advert page shows the outgoing link in plain sight rather than hiding it behind an account.