Sample NIH Summary Statement: Impact Score 55 - Understanding a Low-Scoring Application
SBIR Grant Writers · February 13, 2026
An impact score of 55 represents a fundamentally weak application. At this score level, the issues are not about tweaking language or adding a few experiments - they reflect structural problems with the team, the approach, or the premise itself. Understanding why applications score this poorly helps you avoid the same pitfalls.
Note: This summary statement is entirely fictional. All company names, investigator names, and project details are fabricated for educational purposes.
Download the full sample summary statement:
Download Score 55 Example (PDF)The Application at a Glance
This fictional Phase I SBIR application proposed a machine learning platform for discovery of novel kinase inhibitors targeting treatment-resistant solid tumors. The platform combines graph neural networks for molecular property prediction with a generative chemistry module for de novo compound design.
The Silence in the Resume
The Resume for this application is stark: "The application was not discussed. Written critiques from the assigned reviewers are provided below." While this application technically received a score of 55, the brief Resume tells us reviewers felt there was insufficient merit to warrant panel discussion time. Applications at this level are typically set aside in favor of discussing applications with realistic funding potential.
What 'not discussed' signals: When an application is not discussed (or barely discussed), the assigned reviewers' written critiques represent the only feedback you will receive. There is no panel discussion to provide additional perspective. This makes the written critiques even more important to read carefully.
The Reviewer Scores
Reviewer 1
Reviewer 2
Reviewer 3
The scoring pattern tells the story: When Investigator and Approach both score 5+ across all reviewers, the panel is saying the team cannot execute the proposed work and the plan is not sound. These are the two most difficult weaknesses to overcome because they require fundamental changes - not revisions.
Fundamental Issues Identified
This application illustrates several classic mistakes that lead to very low scores:
Team-science mismatch. The PI had machine learning expertise but no drug development experience. For a drug discovery application, this is fatal. One reviewer noted bluntly that this was "a drug discovery proposal led by someone who has never discovered a drug." No amount of computational sophistication compensates for the absence of medicinal chemistry, pharmacology, and regulatory expertise.
Undifferentiated from existing platforms. Multiple commercial AI drug discovery platforms already exist (the reviewers named several). The application did not explain what makes this specific platform different or better. "We use graph neural networks" is not a differentiator when multiple well-funded companies already use graph neural networks.
Missing essential components. The drug discovery approach lacked ADMET prediction (absorption, distribution, metabolism, excretion, toxicity), selectivity profiling, cell-based assay validation, and a lead optimization strategy. These are not optional components - they are fundamental requirements of any drug discovery program. Their absence suggests unfamiliarity with the drug development process.
No experimental infrastructure. The company had computational resources but no wet lab and no identified external partners for compound synthesis and screening. The entire experimental plan depended on unnamed CROs, making feasibility impossible to evaluate.
In silico-only preliminary data. The computational validation results (AUROC of 0.89) were presented without comparison to baseline models or existing platforms, making them impossible to evaluate. More importantly, no compound had ever been synthesized and tested - the preliminary data were entirely computational.
Can This Be Fixed?
At a score of 55, a standard resubmission is unlikely to succeed. The issues are too fundamental to address with revisions alone. If the applicant wants to pursue this area, they would need to:
- Recruit a medicinal chemist and pharmacologist as co-investigators
- Establish wet lab capability or formalize CRO partnerships with letters of support
- Generate experimental validation data - synthesize and test at least a small set of computationally predicted compounds
- Clearly differentiate the platform from established competitors
- Add ADMET prediction, selectivity profiling, and lead optimization to the approach
This would effectively be a new application rather than a revision of the current one, which is the appropriate path at this score level.
See Other Score Examples
Starting Fresh After a Low Score?
Our experts help you rebuild from the ground up. 41.2% Phase I success rate.
Book Free 30-Min Consultation