Create an application form — mobile-optimized

The average attention span on career pages is under a minute. Whatever is not clearly visible or fillable in that time costs applications — especially among professionals who already have a job and are just "looking".

Three measures measurably lower the drop-off rate: one-per-screen reduces cognitive load (one question, one decision), a visible progress bar shows the end of the tunnel, and autosave allows resuming the next day — without losing data already entered.

Also avoid early hurdles. Forcing account creation before the first question is the most common reason for abandonment. Better: collect data, then optionally create an account or offer a magic link for later status checks. Every second the applicant has to wonder "do I really want this?" loses an application.

A good application form has a narrow required core and generous optional extensions. Required: first and last name, email, desired position, CV upload and a privacy notice with consent. An initial application usually needs nothing more.

Optional but useful: phone number (for callbacks), earliest start date, salary expectation (open-ended, not required), LinkedIn/Xing URL and a cover letter field. Make the last one optional — many top candidates have no patience for boilerplate, a good CV is enough.

Avoid fields like date of birth, marital status, photo or nationality. In Germany they are problematic from an anti-discrimination perspective (AGG) and create bias during screening. If you need this data — e.g. for contracts — ask after the offer.

PDF is the gold standard for application documents — true to layout, platform-independent and not accidentally editable. Restricting to PDF signals professionalism and avoids layout chaos on opening.

In practice, moderate tolerance makes sense: PDF, DOC and DOCX cover most reality. JPG/PNG are legitimate for scanned certificates. Size limit: 10 MB per file and 25 MB per application are generous but not abusable. Server-side MIME type validation (not just checking the extension) prevents smuggled EXE files.

Allowing multiple files per question — such as "certificates" as multi-upload — saves applicants from merging files. Anti-virus scanning before storage is mandatory, ideally via ClamAV or a cloud service. Never store uploads in the web root and serve them only via signed URLs that expire after 24 hours.

Applicant data is particularly sensitive — often including health data (disability status), religious indicators (church-run employers) or family situation. GDPR and BDSG demand correspondingly strict handling.

Retention: maximum 6 months after rejection to cover AGG claim periods (2 months plus buffer). Longer retention only with explicit applicant consent — e.g. for a talent pool. An automatic deletion routine per position (not manual) is mandatory.

Anonymized review — hiding name, photo, date of birth, address — demonstrably reduces bias and is already standard in many public agencies. Access and erasure requests must be answered within 30 days; a self-service portal with magic link saves effort and automatically documents the processing. A DPA with the ATS provider is non-negotiable.

An application form is only the beginning. Real efficiency comes afterwards — when input automatically lands in the right Slack channel, recruiters and hiring managers are notified in parallel and the ATS keeps status in sync.

A lean setup: form fires a webhook to a small endpoint, which forwards to Slack (channel per role), creates a record in the ATS in parallel and sends an automatic confirmation to the applicant. The confirmation should be concrete — "we will get back to you within 5 business days", not "thank you for your application".

For screening, simple scoring pays off: mandatory qualifications are checked via conditional logic already in the form (e.g. "driver`s license available") and automatically flagged. Obvious mismatches land in a separate Slack channel and do not block the main one. Important: never use scoring as the sole decision, always as pre-sorting with human review.