Public debate about funding is almost entirely about the individual application — its quality, its approval rate, its timing. For a budget decision about an application infrastructure, that perspective is too narrow. Anyone using funding programmes strategically submits not one application over three to five years but several — across programmes (ZIM, the research allowance, the EIC, Horizon clusters, DFG individual grants, state programmes), across funding bodies, across consortium partners. The central economic question is therefore not whether a single application is worth the work, but: from which cycle does a structured application platform pay for itself compared with an ad hoc, project-by-project approach.
The most precise publicly available data on this question comes not from Germany but from the R01 regime of the US National Institutes of Health. There, how the probability of success of a second submission — the A1 resubmission — compares with the first has been recorded transparently for decades. The figures do not transfer one-to-one to German or European programmes, but they give an empirically robust indication of the structural leverage of reuse. This piece reconstructs the NIH data, places it against German programme logic, and derives from it when and why a platform architecture for application management makes economic sense.
What resubmission means
In the narrow sense, resubmission means submitting an unfunded application again in a later review cycle of the same programme, revised on the basis of the documented reviewer comments. In the NIH R01 regime the mechanism is formally codified: a first submission is called A0, the revised second submission A1. Since the policy change of 25 January 2009, only one resubmission (A1) is permitted per research idea[5]; any further submission of the same question is only possible as a formal new submission (A0), which has been allowed again since 2014[6].
For the economic view of interest here, the definition has to be drawn more broadly. Resubmission is not only the formal reissue of the same idea in the same programme. It also covers recycling core components — the problem narrative, the state-of-the-art analysis, the description of methods, the presentation of the team, the work breakdown, the financial plan, the impact argument — across different programmes and funding bodies. A technical problem structured for a ZIM cooperation reappears in similar form in EUREKA applications, Horizon Cluster 4 calls and DATIpilot applications. Documenting the first submission so that individual components can be reused modularly measurably lowers the marginal cost of every further application. Treating every submission as a one-off — with no orderly filing, no versioning, no link from comment to change — means paying the full initial cost again with every new cycle.
The NIH R01 data in detail
Through the NIH Data Book and the Research Portfolio Online Reporting Tool (RePORT), the NIH publishes detailed metrics on award rates broken down by application type, submission number and fiscal year[4]. The success rate is the ratio of applications approved to all applications decided in competition in a fiscal year. For R01-equivalent funding — the NIH's main basic funding for individual researchers over three to five years — the overall success rate has been stable between 18 and 22 per cent since the mid-2010s[4]. The figure varies between institutes (NCI, NIAID, NHLBI and others) and between individual submission windows.
The split central to this piece is between A0 and A1 applications. The internal NIH analysis by Lauer and colleagues from 2017 evaluated more than 83,000 unsolicited A0 R01 applications from fiscal years 2012 to 2016 and followed their path through the resubmission system[3]. The finding is unambiguous: the probability of an A1 application being funded was considerably higher than that of an A0 application — and the difference grew with a better initial impact score on the original A0 submission. Those who resubmitted with an impact score in the 10–30 range were funded in around 65 per cent of cases; in the 30–40 range around 55 per cent, in the 40–50 range around 35 per cent[3][2]. At the same time, the majority of applicants with good initial scores did in fact choose to resubmit — around 85 per cent in the 10–30 range, 75 per cent in the 30–40 range — while the ratio drops sharply for weak or undiscussed applications[3].
A second piece of work, concerning behaviour below the funding threshold, is the 2016 study by Boyington, Antman, Patel and Lauer in Academic Medicine[1]. The authors examined 821 unfunded R01 applications from early stage investigators at the National Heart, Lung, and Blood Institute from fiscal years 2010–2012. What was surprising was less the share that was actually resubmitted (51.4 per cent of all applications) than its distribution: applications with a percentile score below 50 were resubmitted in 82.3 per cent of cases, those with a percentile score of 50 or higher in only 34.2 per cent[1]. The only independent statistical predictor of the resubmission decision was the percentile score — not demographic characteristics of the applicants, not institutional factors[1].
In 2023 the NIAID additionally showed that newly submitted applications that had previously failed as an A1 do no worse in fresh competition than average A0 applications[6]. That underlines the core finding: the extra round of submission produces better rates not because reviewers grant a second chance, but because applicants submit substantially revised work on the basis of documented criticism. The reuse lever works through the quality of the second version, not through an administrative bonus.
Cycle 2 as the tipping point
Translate the figures quoted above into a costing and you arrive at a simple finding. Suppose a first submission costs an organisation — counting internal and external effort — a certain effort X for research, building the narrative, coordinating partners, financial planning, formalities. The probability of an A0 submission being approved is around one fifth; the expected grant is therefore one fifth of the funding amount minus X. An A1 submission does not cost X again but a smaller amount X' < X — namely the incremental revision work on the basis of the reviewer comments. In the NIH data, the probability of success rises by a factor of one to three depending on the initial score[2]. The gap between X and X' is where a structured platform applies its leverage: it reduces X' further by making reuse explicit — as versioned components, annotated change deltas and searchable reviewer evidence.
On this calculation, the tipping point at which a platform recovers the cost of building it typically falls at the second or third submission cycle. In the first cycle the initial effort dominates — the platform creates no value beyond what a cleanly organised document drive also delivers. In the second cycle the gap between X and X' becomes visible: narrative components, team descriptions, work package structures and budget models already exist and can be revised specifically against the reviewer comments rather than built from scratch. From the third cycle the structural reuse effect dominates: a narrative on a core technical question that has already been revised twice goes into its third attempt with fewer errors, tighter alignment between the technical and administrative presentation, and a better fit to what reviewers expect.
This logic presupposes that reviewer comments are recorded systematically, attributed, and explicitly answered in the revision. In NIH practice that is institutionalised: every resubmission contains an introduction-to-resubmission document in which the applicants explain, for each individual reviewer point, how they addressed it[5]. That structural requirement is a substantial part of what explains the higher A1 rate: not the act of resubmitting, but the enforced methodical review of the whole chain of argument between the cycles. A platform that represents this review process in machine-readable form — which comment, which change, which new evidence — turns the one-off NIH mechanism into a routine reusable across programmes.
Carrying this over to German programmes
The NIH figures do not transfer one-to-one to German funding programmes. The three main differences: first, most German programmes have no formal A1 category — anyone rejected by ZIM, by a DFG individual grant or by a BMBF thematic programme generally submits not a "revised" application but a new one. Second, the review systems differ: NIH study sections deliver structured reviewer comments in a standardised format from a rotating membership; DFG reviews typically come as free-form prose, and project management agencies' responses are often terser and focused on formal grounds for rejection. Third, the selection mode is often hybrid: two-stage procedures with an outline and a full application (Horizon Europe in some clusters, DATIpilot, the EIC Accelerator) have different reuse properties from single-stage procedures.
Despite these differences the underlying mechanics are comparable. For its individual funding in recent reporting years, the German Research Foundation reports funding rates in the region of a third; its metrics portal documents the distinction between the funding rate (the share of applications approved) and the award rate (the share of the sum approved)[9]. The DFG's guide to dealing with rejected applications explicitly recommends the option of a revised new submission and notes that the reviewers' points are decisive for that revision[11]. The DFG's 2024 annual report gives, for the reporting year, a total approved project sum of over €3.9bn across around 30,940 live projects[10] — the absolute scale at which reuse effects accumulate over the years.
For Horizon Europe the situation is tighter still. The European Commission's interim evaluation of 30 April 2025 (SWD(2025) 110 final) finds that of the 97,403 proposals rated high quality but not funded, 20,890 received a Seal of Excellence in Horizon 2020[7]; the oversubscription against the available budget is of the order of a further €82bn needed to fund all qualifying proposals[7]. In a regime with structurally low success rates, the value of every reusable component rises: every state-of-the-art analysis that does not have to be written again, every consortium matrix that does not have to be rebuilt, is a real cost item.
The European Research Council has even built resubmission directly into its application regime. Under the ERC Work Programme 2026, applications rated A at the first evaluation stage — whether or not they were invited to the second stage — are exempt from any resubmission restriction and can be submitted again immediately or in a later call[8]. A B rating means sitting out one call, a C rating two[8]. This explicit resubmission architecture makes the structural importance of the second cycle a formal part of the programme — and shifts the economic lever even further towards reusability.
What makes a platform reusable
Not every document store is a platform. The difference lies in granularity and in the structure of links. A document drive holds application files as wholes; a platform files the components of an application individually and gives them metadata that makes later reuse possible. Four core requirements follow from the NIH and Horizon findings.
First: modularisation. An application consists of recurring components — the problem narrative, state of the art, description of methods, work packages, the impact section, the team description, the schedule and financial plan. These components have to exist as units of their own with their own version history, not as sections in a single Word document. Only then is it evidenced which version of an impact argument went into which application — and which version contains a revision on the basis of which review.
Second: linking reviews to changes. In the NIH regime every resubmission carries an introduction-to-resubmission document addressing each reviewer point explicitly[5]. The logic behind it — comment, change, supporting evidence — should be represented in a platform in machine-readable form. Anyone who wants to know in cycle 3 which criticism of a description of methods was already answered in cycle 2, and how, has to be able to call those links up — without digging through old email threads or document versions.
Third: tags across programmes. The same technical problem appears in different programmes in different language — ZIM modules, Horizon cluster destinations, DATIpilot themes, state programmes. A platform has to hold these translations as links, so that the same component can be reused in different programmes without being reinvented each time.
Fourth: anchoring independent of individuals. The knowledge work on an application is traditionally tied closely to individual people — a project manager, a researcher, an external consultant. When that person leaves the company, a considerable part of the application knowledge typically leaves with them. A platform that files reasoning, source evidence, responses to reviews and partner agreements independently of individuals removes that knowledge from the effects of turnover. The ZEW's Mannheim Innovation Panel has recorded the innovation activity of German companies annually since 1993 and documents systematically how knowledge is anchored organisationally in innovation processes[12] — structural anchoring outside individual heads is one of the central factors.
Limits of transferability
Three qualifications have to be made explicit when carrying the NIH findings over to German and European programme logic. First, the NIH figures address a specific form of application — basic funding for an individual researcher over three to five years in a structured peer review system. Collaborative projects such as ZIM cooperations, Horizon consortia or EU partnerships have different reuse properties: a larger share of the work goes into assembling the consortium and coordinating partners, a smaller share into individual scientific presentation. A platform's leverage is if anything stronger in collaborative programmes — because it carries partner relationships and role definitions as reusable components — but the NIH percentages do not transfer directly.
Second, reviewer feedback in German programmes is often terser and less structured than in the NIH regime. Without a detailed review, the most important lever of A1 revision is missing — the precise response to criticism you can point to. A platform can only close that gap to a limited extent; but it can capture every available signal — rejection letters, oral feedback from project management agencies, minutes of clarification meetings — in structured form and make reusing them easier.
Third, Germany has no central database comparable to NIH RePORT that publishes success rates broken down by first and subsequent submission. The DFG metrics portal gives programme success rates[9] but does not systematically separate first-time from repeat submissions. For BMBF thematic programmes and the project management agencies, such breakdowns partly exist only internally. That limits the empirical evidence base for German figures. Carrying the NIH evidence over is therefore structural, not numerical: the qualitative mechanism — better quality in the second submission through documented learning from reviews — is plausibly transferable to other structured competitive procedures; the exact rates vary.
Fourth and finally, the reuse lever only applies where the organisation itself carries the application work, or at least actively steers it. Anyone who hands the whole application over to an external consultant who does not file their knowledge in a structured, organisationally accessible form sees no reuse curve: the next cycle starts from zero again, with only the consultant's experience-based routine to show for it. The question of where application knowledge is structurally anchored — with the organisation or with an external service provider — is therefore not only a governance decision but an economic one. It decides who realises the reuse lever.
The NIH R01 data show empirically that the second submission is on average noticeably more successful than the first — and that this advantage is connected with documented learning from reviews, not with a procedural bonus. For German and European funding programmes the number does not transfer one-to-one, but the mechanics do. An application platform becomes economic precisely where it scales those mechanics: by versioning components, keeping review-to-change links retrievable, linking programme translations, and moving knowledge out of individual heads into the organisation. The tipping point does not lie at the first application but in the second or third cycle — that is where the infrastructure pays off that in the first cycle still looks like extra work. upsmart is built for exactly that cycle horizon.
- [1]Boyington JEA, Antman MD, Patel KC, Lauer MS: Towards Independence — Resubmission Rate of Unfunded National Heart, Lung, and Blood Institute R01 Research Grant Applications among Early Stage Investigators. Academic Medicine 91(4): 556–562Association of American Medical Colleges / PMC4811707 · 2016Open source
- [2]Lauer MS: Outcomes of Amended ("A1") Applications (NIH Extramural Nexus, Open Mike, 23 March 2017; the original page nexus.od.nih.gov has been offline since 2025, cited via the Internet Archive)National Institutes of Health, Office of Extramural Research · 2017Open source
- [3]Lauer MS: Resubmissions Revisited — Funded Resubmission Applications and their Initial Peer Review Scores (NIH Extramural Nexus, 17 February 2017; the original page nexus.od.nih.gov has been offline since 2025, cited via the Internet Archive)National Institutes of Health, Office of Extramural Research · 2017Open source
- [4]Success Rates: R01-Equivalent and Research Project Grants (competing applications, awards, success rates by type and submission number)NIH Research Portfolio Online Reporting Tools (RePORT) / NIH Data Book · 2024Open source
- [5]Resubmission Applications — NIH Grants PolicyNational Institutes of Health, Office of Extramural Research · 2023Open source
- [6]How to Approach Application Resubmission Strategy — notes on virtual A2 and on the success rates of newly submitted A0s after an unfunded A1NIH National Institute of Allergy and Infectious Diseases (NIAID) · 2024Open source
- [7]Interim Evaluation of the Horizon Europe Framework Programme, Commission Staff Working Document SWD(2025) 110 final, Brussels, 30 April 2025European Commission, EUR-Lex · 2025Open source
- [8]ERC Work Programme 2026 — European Commission Decision C(2025) 5000, annex to the Horizon Europe work programmeEuropean Research Council Executive Agency · 2025Open source
- [9]Metrics portal — success rates in individual fundingGerman Research Foundation (DFG) · 2024Open source
- [10]Annual report 2024 — tasks and resultsGerman Research Foundation (DFG) · 2025Open source
- [11]Approved or rejected — what now? Procedural guide to individual fundingGerman Research Foundation (DFG) · 2024Open source
- [12]Mannheim Innovation Panel — the annual innovation survey of the German economyLeibniz Centre for European Economic Research (ZEW) on behalf of the BMBF · 2024Open source
The 20/80 asymmetry: why writing the application is the smallest part of the work
The FDP Faculty Burden Survey, the EC's Horizon Europe interim evaluation and the ECA error statistics show where the effort actually lies — and where classic consultants structurally cannot reach.
HITL in the application process: what the Dell'Acqua study at BCG measured
758 consultants, a pre-registered field experiment, +25.1% on speed and +40% on quality — and where the lever finds its limits.
Staff changes in the innovation team: 5 questions about keeping the knowledge
Before the innovation manager leaves: what has to be documented so that applications, evidence and audit trails don't leave with them?
From the analysis into the application.
We show the platform on a real case.