Pharma Bloodbath Part XXI: German employers call 12 of every 16 applications unsuitable

Dear MoreThanCareer.de community,

Somewhere in a GMP plant this morning, somebody is writing a deviation report because a fridge ran 0.4 degrees warm for eleven minutes. Tonight the same person will send out a CV containing the word "passionate" and two different date formats.

Save this one before you next open your CV. You'll want it the day you sit down to "quickly update" the thing and stand up two hours later with a new font and the same content. Comment if you think I'm wrong, and repost it for the colleague who is about to buy a template with a sidebar.

A disclosure, as any inspection report requires: the inspector sells remediation. I rewrite CVs and LinkedIn profiles for a living, next to career coaching (job search strategy, interview preparation, salary negotiation training with tailored scripts) and reverse recruitment. It's all at www.morethancareer.de. That makes me the auditor who hands over the report with one hand and a quote with the other. Check the footnotes accordingly.

On scope: the focus is pharma, biotech and CROs. The experiments come from Germany, Austria, Belgium, Sweden, the UK and the US, and most of what they show holds for a CV in Kraków or a résumé in Chicago.

Twelve of sixteen: the failure rate that gets no investigation

Germany's Institute for Employment Research (IAB) asks employers how their hiring went. In 2022 the average new hire took 9 applications. In 2025 it took 16. Employers judged about 4 of those 16 suitable in principle, which is 28%. For specialist and expert roles the pile was 22 and the suitable share 23%. Companies with 250 or more staff counted 27 applications per hire. IT and natural-science service occupations reached 40.¹

Twelve of sixteen, unsuitable. If an airline lost twelve of sixteen suitcases, somebody would be explaining it on the evening news. In hiring, both sides write a LinkedIn post blaming the other, and the post does rather well.

Two caveats. The IAB figure covers all sectors, and "suitable" is the employer's own verdict on applications its own advert attracted. A Stepstone survey of 308 recruiters in summer 2025 was a little kinder: 40% of applications rated as qualified, about one in seven leading to an interview.² Stepstone sells job adverts, so it has a motive to be gentle about what job adverts bring in.

Pharma adds its own queue. In the US, announced pharmaceutical job cuts came to 15,472 from January to September 2026, against 23,510 in the same months of 2025.³ The industry has slowed the pace at which it removes people. The press office will be along shortly to call that growth.

The pile nearly doubled in three years. The recruiter didn't. I've checked: we still come one to a chair.

The six-second rule: kalter Kaffee, served since 2012

You know the number. Recruiters look at a CV for six seconds. It gets repeated with the confidence of a boiling point, usually by somebody selling a template.

It's kalter Kaffee (cold coffee: stale news). The source is a 2012 study by Ladders, a US job board: thirty recruiters, ten weeks, an eye-tracker, six seconds.⁴ Thirty people. You could fit the entire evidence base into a minibus, and it has been driving career advice for fourteen years.

In 2018 Ladders repeated the exercise and reported 7.4 seconds, in a write-up that doesn't say how many recruiters took part.⁵ In clinical research, an omission like that ends careers. In career advice it starts them. Neither study was peer reviewed, and both measured the first glance that sorts a CV into "maybe" or "no". The legend has since promoted that glance to the whole reading.

In April 2018, StepStone Austria and the research firm MindTake sat 33 HR professionals in front of an eye-tracker. The measured first pass lasted 43 seconds. The same people had estimated that they read for about two minutes.⁶ So recruiters believe they read almost three times as long as they do. I recognise the species. I was the species for 17 years, and I'd have sworn to the two minutes under oath.

The Austrian numbers also show where the 43 seconds go: 22.3 on work experience, 7 of those on the most recent job, 6.1 on education, 4.7 on soft skills and 1.8 on the photo.⁶ The picture you ironed a shirt for got 1.8 seconds. It's a vendor study and a small one. I quote it because it's the only measurement from the German-speaking world I could find.

The peer-reviewed evidence agrees on direction without supplying a stopwatch. A 2023 study followed the eyes of 221 recruiters across 2,043 résumés, and the longer a recruiter looked at the experience section, the likelier the résumé was to move forward.⁷

So half the reading time goes to work experience, and your latest job gets more attention than your entire education. Your hobbies get what's left. I have yet to see anybody hired for "reading, travel and cooking", though I admire the thousands who keep trying.

The industry that files CVs with regulators

This is the part that ought to embarrass us a little. Pharma treats the CV as evidence.

• Good Clinical Practice expects an investigator to prove qualification with an up-to-date CV.⁸
• The EU Clinical Trials Regulation wants a current CV in the application dossier, describing earlier GCP training and trial experience.⁹
• The pharmacovigilance system master file must hold a summary CV of the QPPV, ready for inspectors.¹⁰
• Under § 15 of the German Medicines Act, a Qualified Person needs at least two years of practical work in qualitative and quantitative analysis and other quality testing of medicines.¹¹ Somebody counts those months.

In this industry a CV can be pulled in an inspection. Then the same professionals send a recruiter a document in three fonts, opening with a paragraph in which they describe themselves as a results-oriented team player with a passion for excellence. Passion for excellence is not a GxP term.

Regulators have a five-letter test for whether a record deserves belief. The MHRA spells it out in its data integrity guidance: data should be attributable, legible, contemporaneous, original and accurate. ALCOA.¹² The QA readers have just sat up straighter. For the rest of you: it's what an inspector checks before believing a single number you wrote down.

I've run the CV through it. Five letters, five observations. Each starts with the commodity advice, then the evidence, then what to do. German asks of any claim that it should Hand und Fuß haben (have hand and foot: be well founded). Where a piece of advice has neither, I'll say so, which the commodity version rarely bothers to.

Observation 1, A for attributable: "we" did it, apparently

The commodity advice says "quantify everything". I went looking for an experiment showing that numbers in bullet points raise callbacks. I found none. The advice rests on practitioner consensus, mine included, and you should know that before paying anybody for it.

The nearest real evidence is about substance. Kessler, Low and Sullivan had 72 employers recruiting at the University of Pennsylvania rate 2,880 hypothetical résumés on a ten-point scale of hiring interest. A prestigious internship was worth about one full point, roughly half a grade point. A list of technical skills was worth about zero.¹³

Zero. The section you spent an evening arranging into three tidy columns moved employers about as much as the paper it wasn't printed on. These were elite American graduates, so handle with care. The direction still holds up in my experience: employers pay for what you did and skim past the inventory of what you say you can do.

Pharma CVs fail this letter in two opposite ways. The first is "we". We submitted the MAA. We passed the FDA inspection. "We" is a department, and I can't interview a department. German calls it sich mit fremden Federn schmücken (to adorn yourself with borrowed feathers). "Responsible for the TMF" is the quieter cousin. It tells me the TMF existed, and so did you.

The second failure is the opposite reflex: the German-trained scientist who led a global submission and files it under "supported the team". In my experience that reflex costs more interviews than imperfect English does.

The specific action: for each of your last two roles, write five short lines.

• Your part: lead, contributor or sign-off.
• The object: phase, indication, dosage form or market.
• The scale: sites, patients, batches, countries, budget, headcount.
• The systems, by name: Veeva Vault, Medidata Rave, Argus, TrackWise, SAP.
• The outcome: approval, database lock on time, inspection closed without critical observations.

Then cut the skills cloud down to the tools the advert names. One test per bullet: if it would still be true with your colleague's name at the top of the page, it's a job description. Your colleague is applying too.

Observation 2, L for legible: 43 seconds, and you spent them on a sidebar

The commodity advice says "make it stand out". The evidence says make it readable, and it starts with spelling.

Sterkens and colleagues asked 445 recruiters in Flanders to rate fictitious CVs. Five spelling errors cut the probability of an interview by 18.5 percentage points. Two errors cut it by 7.3. About a third of the penalty came from recruiters concluding that the applicant had lower mental abilities, and another eighth from doubts about conscientiousness.¹⁴ It was a scenario experiment, so no one was really hired. The contempt was real.

German has a hunting term for a self-inflicted blunder: einen Bock schießen (to shoot a buck). Two bucks per page cost seven points, and a third of the damage is a stranger quietly deciding you're a bit dim. The rejection email doesn't mention it. The rejection email says they "decided to pursue other candidates", which is accurate: candidates who can spell pharmacovigilance.

Remember who reads you. This industry breeds Erbsenzähler (pea counters) on purpose, pays them well, and then hands them your CV.

Layout next. Roche tells applicants on its own careers site that its recruiters review each applicant and the attached CV. The same page advises against tables, text boxes, logos, images, columns, headers and footers and unusual section headings, and says a .docx file is parsed most accurately.¹⁵ Das Auge isst mit (the eye eats too) is true of a plate. On a CV the eye wants one column and a job title it recognises.

Which brings me to the two-column template with the skill bars, where good careers go to be misread. I've yet to meet anyone who knows what four dots out of five in "Communication" means. I assume the missing dot is listening.

Now length, where the commodity advice, especially the American kind, says one page. Time for jemandem den Zahn ziehen (to pull someone's tooth: to relieve them of a cherished illusion).

• I don't recommend one-page CVs or résumés at all.
• Two pages are the standard.
• Three are fine for a senior candidate, as long as the content is tailored and each line is relevant to the job advertised.

That is my recommendation from 17 years of reading them, and you should weigh it as exactly that, because I found no credible experiment on length in either direction. The famous claim that two-page CVs are 2.3 times as likely to win comes from a simulation by a résumé-writing firm.¹⁶ What exists is convention: Germany's Federal Employment Agency advises a maximum of two pages,¹⁷ and in the Austrian sample 61% preferred two, with up to four accepted for experienced people.⁶

A one-page CV for someone with fifteen years in drug development is a haiku about a clinical programme. The opposite error is Viel hilft viel (a lot helps a lot), a folk belief pharmacology abandoned some time ago. Page three has to be about this job. If it's about 2004, it's a memoir.

The specific action:

• One column, ordinary headings, contact details in the body of the page, job titles written out in full.
• Print it. Hand it to somebody outside your function, set a timer for 43 seconds, then take the page away.
• Ask what your last job was and what you want next. If they can't answer both, the layout has failed, however handsome it is.
• Have a second person proofread. You can't proofread your own CV for the reason you can't tickle yourself: you know what's coming.

Observation 3, C for contemporaneous: your dates are doing the talking

The commodity advice says "explain your gaps". The experiments on dates are the best in this whole field, and they don't back it.

• Kroft, Lange and Notowidigdo sent 12,054 fictitious résumés to US employers. Callbacks ran at about 7% after one month of unemployment and about 4% after eight.¹⁸
• Eriksson and Rooth found in Sweden that past spells of unemployment stopped mattering once work experience followed them, while a current spell of nine months or more was penalised.¹⁹
• Kristal and colleagues sent 9,022 applications to UK employers. CVs stating each job as years worked, without dates, drew about 15% more callbacks than CVs showing a 2.5-year gap, and about 8% more than CVs with no gap at all. Explaining the gap as time spent raising children did no better than leaving it unexplained.²⁰

So the heartfelt paragraph about your sabbatical was Jacke wie Hose (jacket like trousers: no difference either way). What moved employers was seeing the length of the experience before the calendar.

Now the caveat, because this newsletter doesn't sell miracles. That was the UK, before the pandemic, mostly outside pharma. Germany, Austria and Switzerland expect a lückenloser Lebenslauf (a CV with no unexplained gaps), accounting for each month as if the applicant were a consignment of vaccines. In my judgement a CV without dates would look evasive there, and in GxP roles the dates are part of the evidence. I haven't seen the format tested in DACH.

The specific action: keep the dates and put the duration in front of them. "Senior CRA, 6 years 4 months (03/2018 to 06/2024)." Otherwise the reader does the sum in their head, usually wrongly and rarely in your favour. Recruiters studied many things. Arithmetic was optional.

A gap gets one dated line saying what it was. Mut zur Lücke (courage to leave the gap) beats an essay of apology.

And date your training. ICH E6(R3) has applied in the EU since 23rd July 2025.²¹ "ICH-GCP certified" with no year tells the reader you trained at some point between the Declaration of Helsinki and last week.

Observation 4, O for original: the machine wrote it, and it wrote the other fifteen too

The commodity advice says: paste the advert into a chatbot and let it tailor your CV in thirty seconds. Eine Schnapsidee (a schnapps idea: a plan that sounds brilliant until you're sober).

The research first. It's new, and both studies are preprints from freelance platforms, so read them as early signals. Cui, Dias and Ye studied what happened when a large online labour platform gave applicants an AI tool for cover letters. Before the tool, a letter that matched the job post closely predicted a callback. Afterwards that correlation fell by 51%, and employers shifted their attention to work histories. Time spent editing the AI draft went along with more hiring success.²² Galdin and Silbert report the same pattern on Freelancer.com: employers valued a tailored application before language models arrived and stopped valuing it afterwards.²³

Germans know the mechanism from the history books. Print enough of anything and it buys a little less each week. What's left is Einheitsbrei (uniform mush), served sixteen times per vacancy.

Now my own view from the reading side, which is stricter than the research. Using AI to tune a CV is risky if you don't have the right prompts and don't know how to remove the signs of AI afterwards. Those signs go far beyond vocabulary. Deleting "delve" is the easy part. The architecture gives it away:

• bullets of identical length, each opening with a power verb and closing with "resulting in"
• adjectives that arrive in threes, like a boy band
• a summary paragraph that could sit on top of any CV in the building
• a level of polish that no human with a day job produces

In my experience, most recruiters recognise an application that has been clumsily rewritten by AI very quickly, and most of them reject it right away.

The surveys sound milder. In one, 49% of 1,500 US hiring managers called heavily AI-generated application material a red flag.²⁴ In another, 19.6% of 600 hiring managers said they'd reject a candidate over an AI-generated résumé or cover letter, while 52% accepted AI for proofreading or drafting if the result read as authentic.²⁵ Both come from CV vendors and both record what people say they'd do. People are more gracious in a survey than at an inbox with 200 unread applications. I've met them at the inbox.

I'm hardly the man to tell you to keep your hands off the machine. I gave away a CV prompt on this platform in May. Use AI with care.

The specific action:

• Give it your facts and forbid it to add any. A model asked to "improve" a CV tends to improve the truth first.
• Write the prompt like a brief: the role, the advert, your evidence, the rules it must follow. "Make this better" gets you the CV the other fifteen got.
• Then take the machine out again, structure before words. Vary the bullet lengths, break the verb-and-result template, cut the threes, delete the summary adjectives.
• Read it aloud. Whatever you wouldn't say to a hiring manager's face goes.
• Keep only sentences you can defend when someone decides to dir auf den Zahn fühlen (feel your tooth: grill you). That's what an interview is.

If that sounds like more work than writing it yourself, you've understood the product.

Observation 5, A for accurate: the title and the paper trail

The commodity advice says "sell yourself". Fine, as long as what's sold matches what's delivered.

Titles translated out of German tend to gain altitude somewhere over the Channel. A Referent (roughly a specialist or officer) boards in Frankfurt and lands at Heathrow as a Manager. A Sachbearbeiter (an administrator) arrives as a Specialist, lightly jet-lagged. The reference check will find it. Da kannst du Gift drauf nehmen (you can take poison on it: you can bet on it), a phrase this industry should probably use sparingly.

German-trained candidates also do the reverse and file a global study lead role under "project support". Either way the reader ends up hiring someone they haven't met.

The more interesting evidence concerns the paper trail. Kaas and Manger sent two applications each to 528 German internship adverts. A German-sounding name raised the callback probability by about 14%. Among applications that included reference letters with favourable information about the candidate's personality, the difference disappeared.²⁶

Student internships in 2012, so don't stretch it. But look at what happened. A third party vouched, and the name stopped counting. The candidate's own adjectives had achieved nothing. It's the most German result imaginable: we trust a stamp.

It also explains a habit that puzzles everyone else. Employees here are legally entitled to an Arbeitszeugnis (a written reference from the employer),²⁷ and the German government's own portal for international applicants recommends adding certificates and references.²⁸ For regulated roles the paper trail is the whole point. A Qualified Person's CV gets read with a calculator.¹¹

The specific action:

• Use the title the industry would recognise and put the internal one in brackets if it differs.
• For DACH applications, attach your Arbeitszeugnisse and key certificates. For the UK and the US, leave them out and have referees ready.
• For statutory roles, make the qualifying experience countable: months, activities, products.
• Check each number against something you could produce if asked.

The same inspection in six jurisdictions

The five letters travel. The local customs don't.

• Germany: the Federal Employment Agency tells international applicants that a photo is optional yet still expected by most employers, that two pages are the limit, and that date and place of birth belong on the page while marital status and religion don't.¹⁷ A country with some of Europe's strictest data protection, and its employment agency still suggests telling strangers where you were born.
• Austria: the public employment service says two pages at most and calls a photo usual, with no obligation.²⁹ That may be the most Austrian sentence in this article.
• Switzerland: EURES, the EU's job mobility portal, describes two pages at most, with date of birth, nationality and a photo, plus copies of diplomas and employer references.³⁰ It's less an application than a customs declaration.
• UK: the National Careers Service says to leave out your age, date of birth, marital status and nationality.³¹ The British would like to know nothing about you, and will then ask about your weekend for ten minutes.
• Poland: EURES gives one page, two at most.³² Take the two. The consent clause got its send-off in the issue of 23rd September.
• USA: I couldn't open an official source, so this is from recruiting there. The industry document is called a résumé, it carries no photo and no date of birth, and the one-page doctrine is preached with religious fervour. On the first two points I agree with America.

A German CV with photo, birth date and marital status, sent to Boston, gets passed to Legal with tongs.

On the photo, the evidence is uncomfortable. Weichselbaumer sent identical applications to German employers under three identities. Sandra Bauer got a callback 18.8% of the time. Meryem Öztürk got 13.5%. Meryem Öztürk wearing a headscarf in the photo got 4.2%.³³

German has a proverb for this, and it flatters no one: Was der Bauer nicht kennt, frisst er nicht (what the farmer doesn't know, he won't eat). The experiment measured the farmer.

The other side of the ledger: in a 2022 survey for Indeed, 19% of 400 German HR professionals said they reject applications that come without a photo.³⁴ Vendor data, four years old. So leaving the photo off has a price, and putting it on can have a higher one, depending on a face you didn't choose. I can't turn that into a rule, and I distrust anyone who does.

Who gets which observation

Interns, graduates and anyone needing a visa. Aller Anfang ist schwer (every beginning is hard), and the 2026 edition comes with company. UK employers in the Institute of Student Employers survey counted 140 applications per graduate vacancy in both 2024 and 2025, the highest since the series began in 1991.³⁵ In Germany, Indeed's entry-level postings stood 46% below their 2019 level in July 2026, while postings overall were 8% above it.³⁶ Your letter is A. An industry internship or working-student job goes above your education, because the Kessler experiment suggests that's what employers pay for.¹³ Not on one page, either: a thesis project, two internships and your methods need air. If you need a permit, state your status in words. For orientation, the EU Blue Card in Germany needs a gross salary of €50,700 in 2026, or €45,934.20 in shortage occupations and for graduates within three years of their degree.³⁷

Two years in, up to senior level. Also A, with a second helping. Therapeutic area, phase and market belong in the first line under your name, where the 43 seconds begin.

Executives. Your first reader is usually a researcher at a search firm, building a long list at speed. Page one needs the mandate and its scale: the P&L, the headcount, the sites, the launch or the remediation you answered for. Three pages are fine at your level if each line serves this particular mandate. The trouble starts when page three turns into a guided tour of 2003.

Fifty and older. The largest meta-analysis of hiring experiments puts the penalty for older applicants at roughly 40% fewer positive responses, and finds it worse in Europe than in the US.³⁸ Part XVI covered that in detail. Two things are new. Neumark found that where the first screen was age-blind, older applicants reached interviews at normal rates and then received fewer offers once the room had seen them.³⁹ So trimming dates buys the interview and relocates the problem. Prepare for the room.

Second, 142 Flemish recruiters rated fictitious candidates, and those who showed recent on-the-job training in their field, or willingness for it, were more likely to be chosen at any age. The authors' own caveat: this compensates for age bias without removing it.⁴⁰ So your letter is C. Put dated, recent training near the top and compress everything before about 2010 into two lines. Es ist noch nicht aller Tage Abend (it is not yet the evening of all days: it isn't over), though the market does like to check its watch.

For recruiters and hiring managers

Some of this is Eulen nach Athen tragen (carrying owls to Athens: telling experts what they already know). Some of it apparently isn't.

• If 72% of what arrives is unsuitable by employers' own count,¹ part of the cause sits in the advert. Write the must-haves so that the wrong people can recognise themselves.
• Retire the six-second legend. "I can tell in six seconds" is a confession, and candidates who believe it send you CVs built for a glance.
• Publish your formatting rules, as Roche does.¹⁵ It costs a paragraph and saves your team the retyping.
• Look at the Weichselbaumer numbers,³³ then ask whether your first screen needs to see a photo at all.

Close-out

Most of the five observations you can close yourself in an evening or two. One caution comes from our own rulebook. Chapter 9 of the EU GMP guide wants self-inspections done independently, by someone other than the person who made the product, and then adds a sentence I didn't write: "Independent audits by external experts may also be useful."⁴¹

Well, I would say that. You were warned in the disclosure.

Your turn

A like is a nod across the canteen. A comment helps the next reader, and a repost reaches the friend who is about to spend a weekend choosing a font.

• Run the 43-second test on your CV and report back: what did your reader remember, and what did they get wrong?
• Recruiters, honestly: how long is your first pass, and where do your eyes go first?
• One page, two or three: what did you send, at what level, and what happened?
• Have you received or sent an application that was obviously machine-written? What gave it away?
• Photo on or off for DACH, and what changed after you switched?
• QA colleagues: which ALCOA letter does the average CV in your function fail?

Save this article for your next rewrite, and subscribe if you'd like the next part in your own feed. The Inner Circle email list is at morethancareer.de/#newsletter. Subscribers get 15% off my forthcoming book, Surviving the Pharma Bloodbath: The Ultimate Job Search Guide for Pharma in DACH & US, and 5% off coaching and reverse recruitment programmes. Exclusive career tips for the list will follow in future, and LinkedIn-only readers will miss those. For the series on LinkedIn, follow MoreThanCareer.de: https://www.linkedin.com/company/morethancareer/

CV and LinkedIn rewrites and rebranding, career coaching (job search strategy, interview preparation, salary negotiation training with tailored scripts) and reverse recruitment for pharma, biotech and CRO professionals in DACH, the rest of Europe and the US: https://morethancareer.de/cv-linkedin-rewrite-pharma.html

Prost. Und lasst jemand anderen Korrektur lesen. (Cheers. And let somebody else do the proofreading.)

Sources and references

Primary, official and peer-reviewed:

[1] IAB-Forum, Alexander Kubis, IAB-Stellenerhebung 2/2026: fast drei von vier Bewerbungen sind aus betrieblicher Sicht ungeeignet, 3rd September 2026: iab-forum.de

[3] Challenger, Gray & Christmas, Job Cut Announcement Report September 2026, 1st October 2026: challengergray.com

[7] Pina, Petersheim, Cherian, Lahey, Alexander and Hammond (2023), Using machine learning with eye-tracking data to predict if a recruiter will approve a resume, Machine Learning and Knowledge Extraction 5(3): mdpi.com

[8] ICH E6(R2) Good Clinical Practice, section 4.1.1: database.ich.org

[9] Regulation (EU) No 536/2014 on clinical trials, Annex I, section M, point 65: health.ec.europa.eu

[10] EMA, Guideline on good pharmacovigilance practices, Module II, Pharmacovigilance system master file (Rev 2): ema.europa.eu

[11] § 15 Arzneimittelgesetz (AMG), Sachkenntnis: gesetze-im-internet.de

[12] MHRA, GXP data integrity guidance and definitions, Revision 1, March 2018: assets.publishing.service.gov.uk

[13] Kessler, Low and Sullivan (2019), Incentivized resume rating, American Economic Review 109(11): aeaweb.org

[14] Sterkens, Caers, De Couck, Van Driessche, Geamanu and Baert (2023), Costly mistakes: why and when spelling errors in resumes jeopardise interview chances, PLOS ONE: journals.plos.org

[17] Bundesagentur für Arbeit / ZAV, Application and CV, guide for international applicants, issue 01/2026: arbeitsagentur.de

[18] Kroft, Lange and Notowidigdo (2013), Duration dependence and labor market conditions, Quarterly Journal of Economics, figures via the J-PAL summary: povertyactionlab.org

[19] Eriksson and Rooth (2014), Do employers use unemployment as a sorting criterion when hiring?, American Economic Review 104(3): aeaweb.org

[20] Kristal, Nicks, Gloor and Hauser (2022), Reducing discrimination against job seekers with and without employment gaps, Nature Human Behaviour: nature.com

[21] EMA, ICH E6 Good clinical practice, scientific guideline (effective dates of E6(R3)): ema.europa.eu

[22] Cui, Dias and Ye, Signaling in the age of AI: evidence from cover letters, arXiv 2509.25054, revised 12th November 2025 (preprint): arxiv.org

[23] Galdin and Silbert, Making talk cheap: generative AI and labor market signaling, arXiv 2511.08785, 11th November 2025 (preprint): arxiv.org

[26] Kaas and Manger (2012), Ethnic discrimination in Germany's labour market: a field experiment, German Economic Review 13(1): ideas.repec.org

[27] § 109 Gewerbeordnung (GewO), Zeugnis: gesetze-im-internet.de

[28] Make it in Germany (federal government portal), Application: make-it-in-germany.com

[29] Arbeitsmarktservice Österreich (AMS), Bewerbungsunterlagen: ams.at

[30] EURES, Living and working conditions: Switzerland: eures.europa.eu

[31] National Careers Service (UK), CV sections: nationalcareers.service.gov.uk

[32] EURES, Living and working conditions: Poland: eures.europa.eu

[33] Weichselbaumer, Multiple discrimination against female immigrants wearing headscarves, summary by IZA: newsroom.iza.org

[35] Institute of Student Employers, ISE top 10 stats of 2025, 8th December 2025: ise.org.uk

[37] Make it in Germany, EU Blue Card: make-it-in-germany.com

[38] Lippens, Vermeiren and Baert, The state of hiring discrimination: a meta-analysis of (almost) all recent correspondence experiments, IZA DP 14966 (published in European Economic Review, 2023): docs.iza.org

[39] Neumark (2020), Age discrimination in hiring: evidence from age-blind vs. non-age-blind hiring procedures, NBER working paper 26623: nber.org

[40] Dalle, Wybo, Baert and Verhaest (2024), Age discrimination, apprenticeship training and hiring: evidence from a scenario experiment, IZA DP 17225: docs.iza.org

[41] EudraLex Volume 4, EU GMP guide, Chapter 9, Self inspection: health.ec.europa.eu

Employer's own statement:

[15] Roche, Resume parsing FAQ: careers.roche.com

Vendor and directional (commercially interested, treat as indicative):

[2] Stepstone, Hiring Efficiency study (4,023 job seekers, 308 recruiters, summer 2025), reported by Personalwirtschaft: personalwirtschaft.de

[4] TheLadders, Eye tracking online metacognition: cognitive complexity and recruiter decision making, 2012: bu.edu

[5] HR Dive on the Ladders 2018 eye-tracking update, 8th November 2018: hrdive.com

[6] StepStone Österreich / MindTake, Eyetracking-Studie: so lesen Personalverantwortliche Bewerbungen, 16th October 2018: stepstone.at

[16] CNBC on the ResumeGo study of one-page and two-page résumés, 19th December 2018: cnbc.com

[24] Resume Genius survey of 1,500 US hiring managers, reported by CPA Practice Advisor, 13th August 2026: cpapracticeadvisor.com

[25] TopResume, AI in hiring survey of 600 US hiring managers, May 2025: topresume.com

[34] Indeed Germany survey of 400 HR professionals, spring 2022, reported by Personalwirtschaft: personalwirtschaft.de

[36] Indeed Hiring Lab Germany, Virginia Sondergeld, Berufseinstieg: Stellenangebot sinkt, 23rd September 2026: hiringlab.indeed.com

© 7th October 2026 Andreas Schulz. All rights reserved.