Careers Page Over Job Ads: Winning Applicants Online

The figures are clearer than the debate suggests. Small and medium-sized enterprises make up 99.3 percent (KOFA Study 2/2026, German Economic Institute Cologne) of all companies in Germany and employ 64.3 percent (KOFA Study 2/2026, IW Cologne) of all qualified employees subject to social insurance contributions, close to two thirds. A new analysis of Federal Employment Agency data allows a first look at the skills gap by company size, and it is stark: between July 2024 and June 2025, 72.3 percent (KOFA Study 2/2026, IW Cologne) of all vacancies were advertised by SMEs, and they were short of 281,532 (KOFA Study 2/2026, IW Cologne) skilled workers in total. The largest absolute bottleneck fell on small companies with up to 49 employees: they lacked 177,347 (KOFA Study 2/2026, IW Cologne) skilled workers, while the gap at large companies stood at 111,240 (KOFA Study 2/2026, IW Cologne). Anyone competing for the same people as a corporation with its own HR department can hardly afford to stay invisible as an employer.

The second look shows where it binds. The Federal Employment Agency assesses roughly 1,200 (Federal Employment Agency, skilled worker bottleneck analysis) occupations each year and most recently classified 163 (Federal Employment Agency, skilled worker bottleneck analysis) of them as bottleneck occupations in which vacancies are hard to fill. Another 181 (Federal Employment Agency, skilled worker bottleneck analysis) occupational categories are under observation because they could develop in that direction. More than half of the bottleneck occupations, 91 (Federal Employment Agency, skilled worker bottleneck analysis) of them, sit at skilled worker level, which means classically trained trades rather than the much-quoted graduate shortages. Care and health professions, construction and trades, and IT roles are most affected, along with chefs, professional drivers and childcare educators. At the same time, the Federal Statistical Office reported 648,200 (Federal Statistical Office) registered vacancies for June 2026, up 2.5 percent (Federal Statistical Office) year on year, with around 45.6 million (Federal Statistical Office) people in employment in the first quarter of 2026. The labour market is not empty, then. It is contested. And in a contested market, presentation decides.

The most useful shift in perspective is this: the careers page is not an obligatory page and not an image surface, but a conversion path just like the enquiry path for customers. It has an entry point, a chain of arguments, trust elements, a clear next step and a measurable goal. The difference lies in the audience, and that difference is considerable. A customer asks: can they do what I need? An applicant asks: do I want to spend the next few years there? That is a far more personal decision with a much higher barrier, because it is rarely anonymous and rarely without consequence. Anyone who knows the craft rules that turn visitors into enquiries can apply them directly here. Our article on conversion optimisation for more enquiries describes the mechanics that apply to applications too.

Different content follows from that difference in audience. The applicant is not comparing offers but employers, and they are looking for three things: security, prospects and people. Security means the company exists, pays and will last. Prospects mean the role can lead somewhere. People means they want to see who they will spend eight hours a day with. A service page answers none of these questions, and neither does a job ad in portal format. They are answered on the careers page, or they stay open. That a coherent page concept matters more than isolated measures is central to our web design work.

A surprisingly persistent pattern is the job ad as a PDF attachment, linked from a collection page headed We are hiring. The PDF is convenient because it falls out of the word processor, and it costs something at every turn. It reads badly on a phone, it cannot be linked sensibly, it provides no cleanly indexable page, it cannot carry structured data, it cannot be measured, it has no apply button, and its modification date reveals how long the role has been open. The alternative is unspectacular: one URL per role, as a normal, indexable subpage. That turns a dead document into a landing page that ads, referrals and search engines can point to. How pages are sensibly cut and ordered is described in our article on website structure and information architecture.

The same logic governs how filled roles are handled. A job page without an expiry date becomes a liability: it keeps collecting applications for a role that no longer exists and leaves a stale impression on everyone who gets in touch. What works is a maintained expiry date, a pointer to neighbouring open roles and, when the page disappears, a clean redirect to the careers overview rather than an error page. The technical groundwork for this, from indexability through redirects to markup, is summarised in our technical SEO checklist.

The finish line of the careers page is the application form, and that is where what the page built up is regularly lost. The reason is usually a misunderstanding about the situation: application forms are designed at a desk and used on a sofa in the evening. With smartphone penetration at around 82 percent (Statista) among people aged 16 and over, that is not a fringe case. Anyone who encounters a form with twenty mandatory fields, an address request, a cover letter upload and a registration in that situation will abandon it. Not out of laziness, but because the effort bears no relation to the step involved: they want a conversation first, not a personnel file. Why forms should be thought through on mobile first is explored in our article on mobile-first and responsive design.

The short application reverses the order. It asks only for what is needed for the first callback and collects the rest in conversation. Anything not strictly needed now is optional or dropped. What matters is an exit for those who would rather not type: a direct line with hours when someone genuinely answers. And what matters is honesty afterwards, meaning a note on when a response will come, and then the response. In our own projects (project experience), shorter application paths were as a rule not a quality problem: the applications were not worse, there were simply more conversations. What an entry point that leads to exactly one action looks like is shown in our article on the landing page that converts.

Most companies know what an ad cost and only guess at what it delivered. That is because the wrong things get measured. Clicks, impressions and page views say something about attention and nothing about staffing. The chain that counts is short and uncomfortable: views of the job page, started short applications, submitted applications, candidates invited from those, and hires from those. Only that chain shows whether an ad was expensive or cheap, because an ad with many clicks and zero hires is more expensive than one with few clicks and one hire. How to choose metrics that support a decision rather than merely filling a chart is covered in our article on measuring website success.

Attributing this to a channel works without a tracking apparatus. It is enough to give each ad its own target URL or its own parameter so that views and applications can be separated by source. The simplest method of all helps alongside it: asking, in the form or in the first conversation, how someone became aware of the company. That answer is messy, because people misremember, and it is valuable all the same, because it exposes the share no measurement captures, namely the referral from the workforce. The KOFA study explicitly highlights employee referrals as a successful recruiting strategy for SMEs (KOFA Study 2/2026, IW Cologne). Anyone who does not ask credits their effect, by accident, to whichever ad happened to run last.