Why does my CV keep getting rejected?

One rejection tells you nothing; a series tells you almost everything

A single rejection carries almost no information. You do not know how many people applied, whether an internal candidate was already lined up, whether the budget was frozen midway, or whether somebody with exactly the demanded qualification accepted two hours before your file was opened. No conclusion about your document survives that much unobserved variance. And yet most people redraft their CV immediately after the first no.

A series is different, because exactly two things stay constant across it: you, and the adverts you chose. Everything else rotates. That makes the series the only instrument capable of narrowing the cause — provided you kept it. SECO's guidance for jobseekers on arbeit.swiss is to organise the search systematically from the start and to keep a record of applications sent, procedures under way and responses still outstanding. That is not tidiness advice; it is the precondition for the question having an answer.

Which points at the most common mistake, and it is not a mistake in the CV. It is redrafting after every rejection. Each redraft changes the very variable you were trying to measure, and a series of twenty collapses into twenty single events that mean nothing. If you want to know what is wrong, hold the document deliberately still across a block of ten to fifteen applications and change only one thing: which adverts go into the block.

Your own tally is also biased downwards. With 15.4 percent of adverts duplicating a title and location, and the ten largest advertisers alone accounting for 21.5 percent of everything active, twenty applications are often twelve or thirteen real attempts, several of them to the same recipient. Part of the feeling of being turned down everywhere is that double counting.

The funnel: how much of the visible market is actually open to you

The chart above is the calculation nobody performs before applying for the hundredth time. Start with every active advert. The first cut is the language of the text: 20.2 percent are written in English. That single step costs more than everything after it combined, and it applies before the first line of your CV is read by anyone.

The second cut is the requirement rather than the text. Removing English-written adverts that nevertheless demand German, French or Italian at professional level leaves 16.1 percent. The third cut is seniority: dropping titles that carry a leadership word leaves 10.9 percent, a bigger loss than in any other language, because English-language hiring in Switzerland skews senior.

The fourth cut is contract shape. Explicitly full-time leaves 5.2 percent — a large drop, partly because roughly one advert in ten states no contract type at all and is not counted as full-time here. The fifth cut is time: adverts that satisfy everything above and are also no more than 30 days old come to 2.7 percent. Out of a hundred adverts in a results list, about three are realistic targets for an English-speaking, non-managerial, full-time applicant.

The point is not discouragement, it is attribution. Each of those four conditions is entirely ordinary, and almost every jobseeker carries all four silently. Together they leave a very narrow slice. Twenty applications inside that slice with no interview is a genuine signal about your dossier. Twenty applications sprayed across the whole list is not a signal about anything.

Where language decides, and where it does not

The English-speaking market in Switzerland is real but concentrated. In IT and engineering, 32 percent of adverts are written in English and 55.2 percent require no national language at professional level, so a strong technical CV in English competes on its merits. In banking and insurance the figures are 31.7 percent written in English and 42.8 percent free of a national-language requirement.

Outside those fields the picture reverses quickly. In construction and the trades, 0.4 percent of adverts are written in English and only 21.3 percent ask for no national language at professional level. Health and care sits at 1.2 percent written in English against 37.1 percent free of a professional-level national-language requirement — a gap worth understanding, because it means the work is often open to you while the advert is not readable to you. Retail is at 12.9 percent written in English and 31.5 percent free of the requirement.

So a run of rejections in a field where almost nothing is written in English is not evidence about your CV; it is evidence that you were reading the small part of that field which happens to be published in your language, and competing there against people who could also read the other part. The practical move is not a better summary paragraph. It is applying to German-written adverts whose stated language requirement you actually meet — 44.6 percent of the whole market asks for no national language at professional level, and most of those adverts are not in English.

The same unevenness applies to hours. Part-time exists where shifts are planned: 15.3 percent in health and care, 10.2 percent in retail. It barely exists in IT at 1.3 percent and construction at 0.8 percent. If you need reduced hours in one of the latter, the shortage is structural, and no rewrite reaches it.

If it really is the document: the evidence grid

Suppose you did search inside the right slice and still got nothing. Then it is your documents — but almost never the style, the font or the phrasing. It is that a demanded condition was present in the file yet not findable in it. Testing that needs no tool, just two columns on a sheet of paper.

On the left, write every condition the advert states explicitly, one per row, in the advert's own words. On the right, write the single line of your CV that proves it: function, specialist field, period, scope. An empty cell on the right means you do not meet that condition — a selection problem, not a wording problem. A filled cell that lives on page two, or in the third bullet of a role, means it was not seen on the first pass. That is the document problem, and it is repairable in an evening.

Do this for three adverts you were rejected for. The pattern nearly always surfaces as the same empty row or the same deeply buried row: the permit, the workload, the qualification in its Swiss name, the years in the named specialist field. One advert never reveals the pattern; three almost always do. This is precisely why the series is the instrument and the single event is not.

What counts as proof is narrower than most people assume. "Team player" and "resilient" prove nothing, because they appear in every second dossier and can be checked from none of them. A line that states function, field, period and scope proves itself instantly and without a follow-up question. If you would rather not do the comparison by hand, our CV check performs exactly this pass: one advert, your CV, and the list of stated conditions for which no line in the document exists.

What to change first, ranked by what it returns

The order falls straight out of the funnel. The largest return sits in language, because the first cut is the biggest: 45.3 percent of all adverts require German at professional or native level, and that part of the market opens with the level, never with better phrasing. It is also the most expensive change, measured in months rather than evenings.

Second comes recency, and it is nearly free. Only 13.5 percent of active adverts are a week old or less, and among English-written adverts the median age is 38 days. Setting your search up so that you see new adverts first moves your applications into the part of the process where the shortlist is still open, without changing a line of the dossier. It is the cheapest effective change available and it takes one evening.

Third comes honesty about the title. A leadership word is not a negotiating position, it is a decision taken before you applied, and it covers 18.1 percent of the market overall and 30.1 percent of English-written adverts. Deleting those adverts from your own list costs nothing and raises the share of your applications in which your CV is actually assessed on its content.

Fourth comes the workload question, which for many people is the real one. 5.5 percent part-time across the whole market is a hard number, but it is field-dependent, and treating it as a selection criterion from the outset works far better than raising it as a negotiating point in the interview. In none of these cases is a rejection a verdict on you. Most of the time it is the information that a condition printed in the advert was not met.

What the federal figures count as normal

The State Secretariat for Economic Affairs (SECO) publishes a monthly labour-market report. The July 2026 edition, issued on 6 August 2026, records an unemployment rate of 3.0 per cent and a jobseeker rate of 4.8 per cent. Set the registered jobseekers against the vacancies notified to the RAV in the same month and the ratio is about five to one.

That ratio is explicitly not the number of applicants per job. Only part of the market is notified: 65.9 per cent of the notified vacancies fall under the notification duty, and everything else is reported voluntarily or not at all. It is not a competition rate — it is the only ratio of its kind the Confederation itself publishes, and as an order of magnitude it holds.

The second official figure is defined more tightly and says more about a rejection streak. The exit probability into work is calculated as deregistrations into employment divided by the previous month’s unemployed, times one hundred. For July 2026 that gives 6.3 per cent: roughly one registered unemployed person in sixteen leaves the register in a month because they are starting a job. Twelve rejections in four weeks is not outside that.

Half of all deregistrations are not a job

The same table carries a figure that is rarely quoted and takes the shine off the first one. Of all deregistrations from unemployment in July 2026, 50.6 per cent were into employment and 49.4 per cent fell under the heading "without a job". About half of the people who disappear from the statistics do not disappear into a contract but into training, illness, family work, emigration, retirement, or simply out of registration.

Two things follow for anyone counting rejections. First, leaving the statistics is not a measure of a successful search, so anyone benchmarking themselves against somebody else’s "success rate" is benchmarking against a number that means something else half the time. Second, long-term unemployment is not a fringe case: 16.7 per cent of the unemployed were registered as long-term unemployed in July 2026.

Together those give the sober frame a personal rejection streak belongs in. It is neither proof that something is wrong with you nor a reason to change nothing. It is a stretch of time you can learn from with the right notes — and nothing without them.

Who decided in your series depends on the language you applied in

This is the part only a listings stock can show. On 31 August 2026 we split every live advert by the detected language of the advert text and counted, inside each language part, how much of it the largest advertisers hold. The German-language part is 36.2 per cent of all live adverts, the English-language part 22.2 per cent, the French-language part 10.1 per cent and the Italian-language part 1.7 per cent.

The English-language part is the least concentrated of the four. Its ten largest advertisers hold 17.2 per cent, against 19.2 per cent in German, 35.3 per cent in French and 71.3 per cent in Italian. It also has the longest tail: 14.3 per cent of English-language adverts come from a house with exactly one live advert, against 8.0 per cent in German, and advertisers with ten or more open roles cover only 44.5 per cent of it, against 61.3 per cent in German.

For a reader applying from abroad or newly arrived, that cuts both ways. A streak of rejections in the English-language part is spread across many separate houses, so it carries more information about you than the same streak inside a small language part, where a handful of recipients can account for most of it. But a long tail also means most of those employers are hiring once — there is no second advert to apply to next month, and a good impression cannot be banked.

One word about a neighbouring figure, to avoid confusion: the question above puts the ten largest advertisers at 21.5 per cent of the market. That measures the whole stock, all languages together. The figures here divide that same stock by advert language first, which is why the individual parts sit above and below it. The two measurements do not contradict each other; they do not share a base.

The largest French- and Italian-language advertiser is a staffing agency

The same house sits behind both of the concentrated figures. The largest single advertiser in the French-language and in the Italian-language part of our stock is Pemsa. On pemsa.ch, read on 31 August 2026, it describes itself as an "agence d’intérim" and "agence de placement" for the technical trades of construction, building services, electrical work and industry, with its own branches in Basel, Biel, Fribourg, Bulle, Geneva, Lausanne, Martigny, Bellinzona, Locarno and Lugano.

The adverts are not repetitions of one text: this advertiser’s Italian-language listings carry 66 different job titles across 15 locations in Ticino. Across the whole stock the same house accounts for 2.3 per cent. Its visibility does not come from being large; it comes from being strongly present in the small language parts of the market.

That sits alongside a figure already published elsewhere in this section: 12.0 per cent of live adverts come from an agency or staffing firm and 88.0 per cent directly from the employer. That is the share across the whole stock. The figures here are a different cut through the same reality — not all intermediaries across everything, but one house inside one language part.

What a rejection from an agency means, and what it does not

Anyone who regularly and for payment brings jobseekers and employers together needs an operating licence from the cantonal labour office under the Employment Services Act; hiring out staff requires the same. That is not a detail for lawyers. It is the reason a "no" from that direction means something different from a "no" from the employer.

An agency decides against a mandate, not against a person. If your profile does not fit the assignment currently being filled, it ends there — but your dossier normally stays in the house for the next assignment. A second rejection from the same agency is therefore often not a second assessment of your documents; it is the same assessment applied to another mandate.

In practice: count the recipients in your series, not the applications. Five rejections from three houses are three decisions. And if two of those houses are agencies, the share of your series in which an employer ever saw your dossier is smaller than the number of applications sent suggests.

What actually widens the circle of deciders

We tested what changes about the set of possible recipients when the search area is opened. For each occupational field and canton we counted how many different advertisers published at least one advert in the last thirty days — first in the canton alone, then in the canton together with its neighbouring cantons. Across 55 usable slices the median factor is 2.8.

The effect depends strongly on the field, and that is the real finding. In sales and hospitality the median is 5.8, in commercial and finance roles 5.0, in IT and engineering 4.9. In health and care it is 2.6, in construction and trades 1.7, in education and social work 1.6, in logistics and transport 1.5. Anyone searching in one of those last four gains little from the neighbouring canton — the same or similarly few houses sit there too.

The factor is largest for small cantons next to large ones. A slice in Thurgau, Schaffhausen, Zug or Solothurn opens up several-fold towards Zurich or Aargau; a slice that already contains Zurich gains comparatively little. And an hour of commuting is a real price that none of these numbers contains.

Three fields to record for your next ten rejections

To make a series readable you need three notes per application, and none of them is about your document. First: who received it — the employer, an agency or staffing firm, a public administration? Second: what language was the advert written in? Third: at what point did the "no" arrive — before any contact at all, after a phone call, after an interview?

Those three fields separate cases that feel identical. Ten rejections, all before any contact and all to the same two houses, are two decisions and a selection problem. Ten rejections from ten different employers, four of them after an interview, are ten decisions, and the weight moves from the dossier to the interview.

For the fourth question — whether it is the document, and exactly what in it — the evidence matrix sits on the neighbouring question on this page, and how many applications it takes before an improvement is actually demonstrated is worked through in "How do I measure if my CV improved?". What you can request after a rejection is covered in "What do I do after a rejection?". This page deliberately stops where those begin.

Where these numbers stop

The language shares cover the adverts where we detect a language of the advert text, which is roughly seven in ten live listings. The Italian-language part is small at 1.7 per cent of the stock, and percentages on a small base move more — they describe the structure reliably, but a second decimal place would be invented.

Advertisers are matched by the name as it appears in the advert. Different spellings of one house are merged, subsidiaries advertising under their own name are not, so the published share of the largest advertisers is a lower bound. And anyone applying in more than one language is moving through several of these parts at once.

One thing worth knowing if you read English only: SECO publishes this monthly report in German, French and Italian, and it says so on its own inside cover. The English path on the federal server returns the German file — we downloaded both on 31 August 2026 and they are byte-for-byte identical. The English-language labour-market figures you may find elsewhere are secondary; the primary source is the German, French or Italian edition.