Why 44.7% of EXIST applications get rejected (and what makes the 55% that pass)

More than half of the rejections cite this. The pattern: founders treat "new for the German market" as innovation. Reviewers want "new technically or methodologically, with citations." If your three reference papers are from your own group, you do not have an Innovationshöhe argument; you have a thesis topic.

EXIST is a Hochschul-Antrag. You need an academic mentor (typically a professor) who has signed off on the project and is willing to be named. Not a friend at the university, not a postdoc you met once - a named, willing supervisor. Applications without one get auto-rejected before content review.

Reviewers read fifty applications a month. They have learned what ChatGPT-cadence looks like - safety-rounded conclusions, "moreover, furthermore" connective tissue, abstract-noun-heavy paragraphs. Use the LLM to escape the blank page; do not submit its output. The 2024 monitoring report does not name AI use as a rejection reason, but five PTJ reviewers we spoke to did.

Underrated. A 30-page PDF with broken numbering, inconsistent heading styles, and figures pasted in at 72dpi reads as low-effort. Reviewers do not call it a rejection reason out loud - they call it "the application did not convince." Spend the last evening of week four on formatting.

Reviewers know fresh teams have not closed sales. They want evidence you have spoken to potential customers, not that you have signed contracts. Three named conversations, with what each customer said, is the bar. Zero conversations is rejection-territory.

EXIST funds product startups, not consultancies. If your business model in the application reads as "sell my expertise by the hour," the application gets refused regardless of merit. Reframe as a product or platform play even if the early revenue is service-shaped.