Here's a number that should haunt every founder who's ever thought about SBIR funding: the average success rate across all agencies is roughly 15-20%. That means 4 out of 5 proposals lose. Not because the technology is bad — but because the proposal is.

I've read winning SBIR proposals and losing ones. The technology gap between them is often razor-thin. The writing gap is a canyon. Winners tell a story the reviewer can follow. Losers dump technical specs into a blender and hit "submit."

Let's make sure yours wins.

Before You Write a Single Word

The best SBIR proposals are won before the writing starts. Here's what separates companies that consistently win from those that spray and pray.

1. Pick the right topic.

Every SBIR solicitation contains dozens of topics — specific problems an agency wants solved. Most first-time applicants scan the list, find something that vaguely matches their technology, and start writing. That's backwards.

Instead, look for topics where:

• Your technology is a direct answer to the stated need — not a stretch, not a pivot, not "well, we could adapt it"
• You have existing work, data, or a prototype that demonstrates feasibility — reviewers want evidence, not promises
• The topic hasn't been funded repeatedly by the same companies — check award databases on SBIR.gov to see who's won before

Think of topic selection like choosing which mountain to climb. A skilled mountaineer doesn't pick the tallest peak — they pick the one their gear and experience are built for. The same technology can be a perfect fit for one topic and a weak match for another.

2. Talk to the program manager.

This is the single most underused advantage in SBIR. Most agencies allow (and encourage) pre-submission communication with the Topic Author or Program Manager. These conversations are not sales pitches — they're intelligence gathering.

Ask:

• "What would a strong Phase I demonstration look like for this topic?"
• "Are there specific technical approaches you've seen that didn't work?"
• "How does this topic connect to the agency's broader priorities?"

You're not asking them to reveal the scoring criteria. You're learning what matters to the person whose mission your technology would serve. That context shows up in proposals that win.

The Anatomy of a Winning Proposal

SBIR proposals follow a standard structure, but within that structure, the quality of execution varies wildly. Here's what each section needs to accomplish.

Technical Approach (This Is Where Proposals Are Won or Lost)

Reviewers read dozens of proposals per topic. They're technical experts, but they're also exhausted. Your technical approach needs to be clear enough that a tired reviewer at 10pm can follow your logic without re-reading paragraphs.

What winners do:

• Open with a crisp problem statement that mirrors the solicitation language — show the reviewer you understand exactly what they asked for
• Present the approach as a logical sequence: "We will do X, which will tell us Y, which enables Z." Cause and effect, not a feature list
• Include preliminary data — even early-stage results dramatically increase credibility. A graph from a bench test beats a page of theoretical projections
• Be specific about methods. "We will use transfer learning with a pre-trained ResNet-50 model fine-tuned on 10,000 labeled satellite images" beats "We will apply state-of-the-art deep learning techniques"

What losers do:

• Spend two pages on background the reviewer already knows (they wrote the topic — they know the problem exists)
• Describe their product instead of their research plan — SBIR Phase I is a feasibility study, not a product launch
• Use jargon as a substitute for clarity
• Promise results without explaining the methodology that produces them

Innovation and Commercialization

The innovation section answers: "Why hasn't someone already solved this?" The commercialization section answers: "If this works, who pays for it?"

For innovation, be specific about what's new. "No existing solution combines edge computing with federated learning for DoD tactical networks" is a claim you can defend. "Our AI is innovative" is not.

For commercialization, show that you've thought about the business — not just the science. Identify specific customers (government and commercial), estimate the addressable market, and describe your path to Phase III. Agencies want to fund technology that will actually get deployed, not research that dies on a shelf.

A NASA reviewer once told me: "I can forgive a proposal with an aggressive technical timeline. I can't forgive one with no commercialization plan. We're not a university — we fund things that need to fly."

Team and Facilities

This section is about credibility. The reviewer needs to believe your team can execute what you've proposed.

• Highlight relevant experience, not comprehensive resumes. Your CTO's PhD in materials science matters. Their undergraduate GPA does not.
• If you're a small team, that's fine — but address it directly. "Our three-person core team will be supplemented by Dr. [Name] at [University] as a consultant for [specific capability]."
• Describe any specialized facilities, equipment, or data access that gives you an advantage.

The Five Mistakes That Kill SBIR Proposals

1. Not following the format. If the solicitation says 20-page limit with 1-inch margins and 11-point font, then it's 20 pages with 1-inch margins and 11-point font. Proposals that violate formatting requirements get rejected without review.
2. Solving a problem the agency didn't ask about. Your technology might be brilliant, but if it doesn't address the specific topic, it scores zero. Answer the question that was asked.
3. No preliminary data. Even a Phase I feasibility study benefits enormously from showing you've already started. A simulation, a bench test, a pilot study — anything that says "this isn't just a hypothesis."
4. Vague milestones. "We will develop an AI model" is not a milestone. "By Month 4, we will achieve >85% classification accuracy on the benchmark dataset" is a milestone. Give reviewers something they can evaluate.
5. Ignoring Phase III. Agencies increasingly weight commercialization in their scoring. A proposal with strong technical approach but weak commercialization will lose to a slightly less technical proposal with a clear path to deployment and revenue.

After You Submit

SBIR review cycles take 3-6 months depending on the agency. During this time:

• Don't stop working. Continue developing your technology and building relationships with the agency.
• If you're not selected, request a debrief. Most agencies provide written feedback on unsuccessful proposals. This feedback is gold — it tells you exactly what to fix for next time.
• If you are selected, start planning Phase II immediately. The strongest Phase II proposals reference Phase I results that exceeded expectations.

Writing a winning SBIR proposal is a skill, not a talent. It improves with practice, feedback, and understanding what reviewers actually look for. The technology gets you in the conversation. The proposal wins the award.