I have evaluated more than 3,700 companies and proposals as a Verified Expert Evaluator, for grant schemes, accelerators and investment programmes. The rules change from one call to the next. The mistakes do not. What follows is not an official rulebook for any programme: it is my own experience of what costs applicants points, written so you can check your application against it before you press submit. Each point is a pattern I have seen many times, never a single case.
One thing to keep in mind as you read. Evaluators score what is on the page, against a written rubric, often with a time limit per proposal. We are not allowed to fill gaps with what you probably meant, to search for what you forgot to attach, or to reward potential you did not describe. Almost every mistake below comes from forgetting that.
Before you write a word
Answering the form against the call
The simplest and most painful mistake. If a call presents itself as a stepping stone to a larger programme, and the form asks whether you plan to apply to that programme, answering "no" tells the evaluator you do not fit the call. The same goes for any yes/no or dropdown question that mirrors the call's purpose: target group, sector, stage, intended next step. These fields are short, so applicants fill them in last and fast. Evaluators read them first.
Fix: before writing, copy the call's stated objectives into a list, then check every closed answer in the form against it. If an honest answer is "no", ask yourself whether this is the right call for you.
Not checking eligibility first
Company age, legal form, country of registration, how much funding you have already raised, who owns the shares, which activities are excluded. Each of these can end an application before anyone reads the technology. Exclusions are easy to miss: some programmes fund only purely civilian applications, others exclude certain sectors entirely.
Fix: read the eligibility section of the call and its guidelines before anything else, tick each criterion against your documents, and ask the programme's helpdesk when something is unclear. It costs an email.
Writing a pitch instead of answering the rubric
Most programmes publish the criteria evaluators use, often with the definition of each score. Many applicants never open that document. They write one good company story and paste it into every box, so the evaluator has to hunt for the answer to each criterion. When we cannot find it, we cannot score it.
Fix: structure each section around its criterion, using the rubric's own words as signposts. If the rubric asks for "the team's experience in the sector", write a sentence that starts there.
The documents
Mentioning documents you did not attach
"See the CV of our CTO" with no CV attached. A letter of intent cited in the text and missing from the annexes. A file named "Document1.pdf" that turns out to be the key evidence. A missing CV can cost points on two or three criteria at once, because team, experience and leadership are often scored separately.
Fix: on the last day, make a list of every document the text refers to and tick it off against the uploads. Name files so an evaluator knows what they are: "CV_Name_CTO.pdf", "LOI_CustomerName.pdf".
Numbers that disagree between documents
Revenue forecasts that differ between the deck and the form. A team of six in one place and four in another. A market size that changes from section to section. Each inconsistency on its own is small. Together they tell the evaluator that nobody on the team owns the numbers, and every other claim becomes harder to believe.
Fix: keep one sheet with your key figures (team, revenue, customers, funding raised, market size, timeline) and copy from it. Then have one person read everything in one sitting, looking only for numbers.
Jargon and acronyms
Evaluators read many proposals in a row, often outside their own specialty. Acronyms, internal product names and dense technical language make us work harder to give you points, and tired evaluators give the benefit of the doubt less often.
Fix: explain each technical idea once, in a sentence a smart non-specialist would follow, and spell out every acronym the first time. Ask someone outside your field to read one page and tell you what they understood.
Letting AI write the proposal
Using AI to fix grammar or tighten a paragraph is fine, and many programmes explicitly allow it. Letting it write the substance is another story. Generic text that could describe any company scores low on its own, because it says nothing specific about your technology, your customers or your team. Some programmes also screen submissions with AI-detection tools.
Fix: write the first draft yourself, with your real numbers and examples, and use AI only to edit. Check the call's rules on AI use, and never paste confidential material into a tool you do not control.
The substance
Describing the technology with adjectives, not results
"Cutting-edge", "proprietary", "AI-powered", "revolutionary". These words cost nothing and prove nothing. What scores is evidence: what you tested, on what data or in what conditions, with what result, compared with what alternative. Results on synthetic or internal data count, but as early evidence, and you should say so.
Fix: for each technical claim, add one line of evidence: the metric, the test conditions and the benchmark. If you have no results yet, say what you will measure and when. An honest gap scores better than an unsupported claim.
Claiming a maturity level you cannot show
Many programmes use technology readiness levels (TRL) or a similar scale. Applicants often claim the level they are aiming for rather than the one they have reached. The evaluator then compares the claim with the evidence, and the gap costs you credibility on everything else.
Fix: state the level you have reached, then say what proves it: a prototype tested in the lab, a pilot with a named user, a system in daily use. A modest, well-supported claim is stronger than an ambitious one.
A market that starts from a billion
"The global market is worth 40 billion" tells the evaluator nothing about your business. Top-down figures from generic reports are the most common market section I read, and the weakest. What convinces is a bottom-up estimate: how many customers of a specific kind, paying how much, reachable through which channel, plus any proof that real customers care.
Fix: build the market from the customer up, show the arithmetic, and name what you have: pilots, paying customers, letters of intent, even signed meeting notes. If your plan is to grow in a particular region, name at least one customer or partner there.
Saying nothing about intellectual property
No patent is not a problem in itself. Silence is. Evaluators need to understand why a competitor cannot copy you next year. Trade secrets, know-how, data you own, software copyright, speed of execution: all of these can be a credible strategy if you explain the reasoning and how you protect them.
Fix: write a short paragraph on what you protect, how (patent filed or planned, trade secret, contracts, IP assignment from founders and staff), and why that choice fits your business. Mention whether you have checked that you are free to operate.
A team section that hides who actually works on it
A long list of names with impressive titles, no indication of who works full-time, and no one covering sales, finance or operations. Evaluators look for a team that can deliver the plan, not just invent the technology. Advisers help, but only if their role is clear.
Fix: for each person, say their role, their time commitment and the one experience most relevant to this plan. Name the gaps yourself and how you will fill them, with a hire, an adviser with a defined role, or a partner. Naming a gap scores better than an evaluator finding it.
The video and the ethics section
A video that runs long or says nothing
When a programme sets a time limit for the video, evaluators are often told to stop watching at the limit. Whatever comes after does not exist. A polished brand film is also not a pitch: if we do not see who you are, what problem you solve and the product working, the production budget does not help.
Fix: stay under the limit, open with the problem and the team, show the product or prototype in use, and end with what you need. A phone, good light and a quiet room are enough. Content beats production.
Ticking "no" on every ethics question
Health data, personal data, employee monitoring, minors, profiling, animal testing, possible military use. If your product touches any of these and the ethics self-assessment says "none", the evaluator will flag it anyway, and now it also looks like you did not think about it.
Fix: answer the ethics questions honestly and add one or two sentences on how you handle each issue: consent, anonymisation, data stored in the EU, an ethics adviser, a clear civilian-only scope. It turns a red flag into a sign of maturity.
The pattern behind all 14: evaluators reward applications that are easy to score. Every answer where we expect it, every claim with its evidence, every number the same everywhere. None of this needs money. It needs one slow, critical read before you submit, ideally by someone who did not write it.
The checklist at a glance
| Mistake | What the evaluator sees | Cheap fix |
|---|---|---|
| Closed answers that contradict the call | An applicant who does not fit the call | Check every yes/no and dropdown against the call's objectives |
| Eligibility not checked | An application that may be rejected before scoring | Tick each criterion against your documents; email the helpdesk |
| Pitch instead of rubric | Answers hard to find, so hard to score | Use the rubric's words as section signposts |
| Missing attachments | Claims with no proof, on several criteria | List every document cited; name files clearly |
| Inconsistent numbers | A team that does not own its numbers | One master sheet of key figures |
| Jargon | Extra effort to award points | Explain once, spell out acronyms |
| AI-written text | Generic claims that fit any company | Write the draft yourself; AI only to edit |
| Adjectives, not results | Unproven technology | One metric, condition and benchmark per claim |
| Overstated maturity | A credibility gap | Claim the level you can prove |
| Top-down market | No real customers in sight | Bottom-up estimate plus named pilots or letters |
| Silence on IP | Nothing stops a copycat | One paragraph: what, how and why |
| Vague team | No one to deliver the business plan | Role, time and key experience per person; name the gaps |
| Long or empty video | Content cut at the limit, no product shown | Under the limit; problem, team, product, ask |
| Ethics left blank | An issue the team did not notice | Flag it honestly and say how you handle it |
If you are applying to the EIC Accelerator
Every point above applies, with higher stakes: after three unsuccessful submissions you cannot apply again under Horizon Europe, so a sloppy attempt costs more than the time you spent on it. For the specifics, read how EIC Accelerator evaluators score a proposal and, for the strategic reasons applications fail rather than the avoidable ones, 5 reasons EU grant applications fail.
If you want an outside read before you submit, the fixed-price EIC Fit Assessment goes through your project the way an evaluator would, in one week. If you would rather start on your own, the free EIC readiness check takes a few minutes.
Questions
What is the most common avoidable mistake in funding applications?
In my experience, inconsistency: answers in the form that contradict the call's purpose, and numbers that differ between the form, the deck and the video. Both are free to fix with one careful read before submitting.
Do evaluators look at documents I forgot to attach?
No. Evaluators score what is submitted. If the text refers to a CV, a letter of intent or a test report that is not attached, the claim counts as unsupported, and it can cost points on several criteria at once.
Can I use AI to write my grant application?
Check the rules of the specific call. Many programmes accept AI for editing, but the substance should be yours: generic AI text scores low because it says nothing specific about your company, and some programmes screen submissions with AI-detection tools.
Do I need a patent to score well on intellectual property?
No. Evaluators want to understand why a competitor cannot copy you. A clear strategy based on trade secrets, know-how, data or software copyright, with the reasoning and the protection measures explained, can score well. Saying nothing scores poorly.
What should a pitch video for a funding application show?
Who you are, the problem, the product or prototype working and what you need, within the time limit. Evaluators are often told to stop at the limit, and a brand film without the team or the product does not help your score.
Verified Expert Evaluator · 3,700+ companies and proposals reviewed
Want an evaluator to read it before you submit?
The fixed-price EIC Fit Assessment checks your project against the criteria in one week and tells you what to fix first. If you want to talk it through, book a call.
Book a discovery call →