What does an ATS score of 80% mean — and how seriously should you take it?

Where the number comes from: four checkers, as at 4 September 2026

Every cell is taken from the vendor’s own public pages, fetched on 4 September 2026, or for the first column from the running product. Where a vendor page does not say, the cell reads "not stated" — never a guessed No.

What the number costs, as at 4 September 2026

How alike are two Swiss adverts for the same job? 383 adverts, 4 September 2026

Read as: of the 20 most frequent content terms in one advert, a median of 15% (German) to 40% (English) appear anywhere in the text of a second advert for the same occupation at a different employer. Whole adverts were compared, not CVs, so these are generous upper bounds.

Who calculates the score — and who certainly does not

The employer does not send you a percentage. What an applicant tracking system does with your dossier is invisible from outside, and no CV-checking vendor has access to it. Every "ATS score" you can obtain online is a tool marking your CV against its own rulebook. That is not an accusation; it is simply the only vantage point a tool has.

You can see the rulebooks in the vendors’ own words. Jobscan states on its home page that the match rate rests on five priorities in this order: hard skills; education level, and expressly only when the advert names an advanced degree; job title; soft skills; other keywords. Then comes the sentence that matters most for a Swiss dossier — resume word count and measurable results are not factored into the match rate. If the strength of your CV is the results you can prove, the percentage does not see them.

Resume Worded, by contrast, advertises that the same resume always gets the same score. That is honestly put and genuinely useful for measuring your own progress. It also means the number cannot, by construction, know anything about a specific job: a figure that does not move when the advert changes is not measuring the advert.

And at cvswiss.ch, the Swiss vendor in this field, the free check takes a file and nothing else. Its home page shows a specimen CV labelled "ATS Score: 95%" with no advert anywhere in the demonstration. That is internally consistent — a score without an advert is a statement about form and readability, not about fit.

80% is already above the vendor’s own recommendation

The common assumption is that higher is always better. The vendor that popularised the metric disagrees. In its own FAQ, under "What is a good ATS score for a resume?", it recommends a match rate of 75% and notes that many career counsellors and users see success with just 65%. Immediately after that comes the warning: there is such a thing as an over-optimised resume, and it might not be possible to score above 75% without overstuffing your resume with keywords.

Now do the arithmetic. If a tool extracts twenty terms from the advert, the display can only move in steps of five: 15 of 20 is 75%, 16 of 20 is 80%. The entire distance between the vendor’s recommendation and your 80 is one word. Which is why pushing an 80 to an 85 is a poor trade — you buy two words and you pay in sentences that a hiring manager reads as a list of buzzwords.

The second reason 80 is not a seal of approval is that the systems behind Swiss adverts are not the ones these scores were calibrated against. Jobscan names as its basis "top ATS including iCIMS, Recruiterbox, Lever, Greenhouse and Taleo". On our own reading of the Swiss adverts whose apply link resolves to a known system, Greenhouse accounts for 8.4% and Lever for 3.8% — about 12% between them. The two largest are Workday at 28.8% and a Swiss product at 21.6% that no English-language ATS guide mentions.

We measured it: two adverts for the same job

If a score measures how closely a text matches an advert, that idea can be stress-tested directly — with no CV and no guesswork. Instead of a dossier, take a whole second advert for the same occupation and count how many of the first advert’s leading terms appear in it at all. A full advert is far wordier than any CV, so whatever comes out is a generous upper bound on what a real applicant could score.

Measured on 4 September 2026 across 383 live adverts, one per employer and no apprenticeships: for English-language Swiss adverts, the median overlap between two adverts for the same occupation is 40%, across 722 pairs from different employers. For two adverts in unrelated occupations it is 30%. So the distance between "the same job" and "any job at all" is ten points, and most of any such number is shared advert language rather than the role.

English is the best case of the four languages, and it is still 40%. Only 2.6% of same-occupation English pairs reached the 75% Jobscan recommends — and those were two complete adverts, not a two-page CV. If you have been treating a score below 90 as a personal failing, that is the correction: the ceiling is set by how differently two employers describe the same work.

As a check on method we repeated the whole calculation with a stricter rule for choosing an advert’s leading terms, one that first discards words common across the corpus. The absolute numbers fall — a median of 5% (German) to 15% (English) — but the ordering between languages and the gap between same and different occupations hold. The truth sits between the two rules, and both say the same thing.

If you are applying from abroad, this is the part to read twice

Roughly a third of the people reading these pages are searching from outside Switzerland, and a Swiss employer will read your dossier against Swiss conventions whether or not the advert says so. A photo belongs on the page here, where guides written elsewhere tell you to leave it off. Languages carry a European framework level beside them, B2 or C1, and never the word "fluent". Workload appears as a Pensum in per cent, so a 60–100% line in an advert is a real requirement. Two pages is the norm. And a Swiss dossier normally carries Arbeitszeugnisse, the written employer references every previous employer must issue on request; nothing in a US-calibrated resume score models them.

The permit line is the one that costs interviews silently. If you hold an EU or EFTA passport, or a B, C or L permit, put it on the CV in one line near your contact details. Jobscan’s published list of priorities contains no such field, and Resume Worded does not publish its checks at all — which is precisely why a high score from either tells you nothing about the question a Swiss recruiter asks first.

There is also a split inside the country worth knowing about. English-language Swiss adverts are written in the shape of a US posting: long, keyword-dense, results-oriented. German-language ones are half that length and built out of compound nouns. Most people applying from abroad send one CV to both kinds of employer. The imported ATS advice fits the first and is structurally wrong for the second.

The missing 20% is the actual question

A percentage is an average, and an average hides its extremes. Here the headline number is the equal-weight mean of seven dimensions. Four dimensions at 100 and three at just over 50 makes 80. Six dimensions at 93 and one at 2 also makes 80. Those two dossiers share a number and share no problem at all: the first needs polish everywhere, the second is missing an entire category — any trace of motivation for this particular job, say.

So the useful question is never "is 80 good?" but "which dimension is dragging?". If the keyword score is low, the answer is vocabulary: describe your experience in the advert’s words wherever they are honestly true of you. If the outcomes score is low, the answer is numbers: "responsible for purchasing" becomes "ran a CHF 4m purchasing budget across six suppliers". If personality and environment fit is low, the answer is level: a dossier written for a twelve-person team reads differently from one written for a bank.

And if all seven look reasonable and the number is still 80, that is not a defect. It is the correct answer: the CV fits well and not perfectly, because nobody fits a job perfectly. On this tool 80% is where the label "Strong fit" begins — the bottom edge of the top band, not its middle.

What a percentage structurally cannot do: yes-or-no criteria

The Swiss conditions that actually end applications are not scales. You either hold a residence permit or you do not. The qualification is either the one named or it is not. German at C1 required against B1 held is not eighty per cent compliance; in many processes it is the end of the process. An average would net such a criterion off against well-met ones and destroy exactly the information you ran the check for.

That is why the three most consequential checks here sit outside the percentage: permit and eligibility, languages with their level, and the photo per Swiss convention. Each is graded have, partial or missing and shown on its own. A second fixed list covers machine readability with nine elements — contact details as real text, conventional section headings, a single-column layout, the job title, consistent date formats, an extractable text layer, length, keyword coverage and the photo — again graded individually rather than averaged into a number.

That separation is the whole point of the exercise. An 80 with a clean gate is a good dossier that needs polish. An 80 with no permit line is a dossier that triggers a question you could have answered in one line.

What to actually do with an 80

First, check whether an advert was involved at all. If the tool only took your file, the number is a mark for form, and it says nothing about whether you fit this job. Second, read the individual scores rather than the mean, and work on the lowest dimension rather than on the headline. Third, deal with the yes-or-no items first, because no amount of percentage can offset them.

Fourth — the step most people skip — check against more than one advert. If the same dossier scores very differently against three adverts for the same occupation, you have learned that the spread comes from the adverts, and you can stop rebuilding your CV for every point. Our measurement predicts exactly that spread, and it is larger in German and French than in English.

Fifth, if you are rewriting anyway, write for the person who reads it at the end. The requirement Swiss adverts name most often is a language with a level; the second most often is a specific qualification. Both belong in one clear line, not in a cloud of keywords.

Who should use something else

If you are applying in the United States, Jobscan is the better fit. Its metric is calibrated for that market, the systems it names are common there, and the free tier of five scans a month covers a small campaign. If you want a pure before-and-after measurement on an English CV, with no advert and no Swiss criteria, Resume Worded is good at it — the stability of its score is an advantage there, not a flaw.

And if you simply want some number about your CV in five minutes, with no account and no payment, the free cvswiss.ch check is the shortest route to one. It really is free and it really does not ask you to sign up. It just cannot answer the question in the title, because it does not read an advert — and the vendor makes no secret of that.

For everyone else — people applying to Swiss adverts who want to know what is wrong with this particular application — a broken-out check against the advert’s own text is more useful than a single figure. Not because our number is better, but because what it is made of is printed next to it.