What SBIR Reviewers Actually Score (and How to Self-Grade Before They Do)

You can read your own proposal a hundred times and not see what a reviewer sees in one pass. You are reading for whether the technology is good. They are reading against a rubric, scoring section by section, looking for the reason to put you in the "no" pile so they can get through the other thirty-nine proposals.

The fix is not to read it again. It is to read it the way they do.

What they actually score

Most SBIR rubrics reduce to seven questions:

1. Soundness and feasibility for this team, this budget, this schedule. Not "is this possible." "Can these specific people deliver this specific thing for this money in this time."
2. Alignment to every objective. Every objective in the topic is a promise they check you kept. One unanswered objective is a hole they will find.
3. Phase I-equivalent evidence. Demonstrated, dated, named. Not asserted.
4. Concrete, measurable milestones with acceptance and dollars. The milestone table.
5. A justified, credible budget. The cost volume.
6. Commercialization realism. Specific market, named customers, real demand evidence.
7. Team and PI qualifications. Including that the PI's level of effort meets the floor.

Every section of your proposal should be visibly written to one of these. If a section does not map to a scored criterion, ask why it is taking up your page budget.

Your trust budget

Here is the mental model that changes how you write: every unsupported claim spends trust; every piece of evidence (a number, a date, a customer, an artifact) replenishes it. A proposal that runs out of credibility by section three does not get a generous read in section six. This is why the feasibility evidence goes early and concrete: you are buying the right to be believed later.

Reconstruct the rubric, then grade yourself

From the criteria above and the specific language of your solicitation, write the scoring sheet the panel will use. Then score your own draft against it, honestly, 0 to 5 per criterion. You are not looking for the average. You are looking for the single weakest section and the specific reason it loses points there, because that is exactly where the panel will put you in the "no" pile.

Then ask the question the hostile reviewer asks: what do they attack first? Usually it is the obvious soft spot: a key hire not yet on staff, a budget number with no basis, a commercialization claim with no customer. Find it before they do and either fix it or pre-empt it in the open (a named role with a letter of intent, surfaced again in the risk register, is a disclosed managed risk instead of a discovered gap).

A free Reviewer Scorecard

We built this into a tool you can use right now, no signup: paste a draft, get a per-criterion score against the reconstructed rubric, your weakest section named, the specific reason it loses points there, and the fix. It is the lowest-friction way to find out whether you are about to spend three weeks on a proposal that loses for a fixable reason.

A good score means your proposal is consistent, complete, and free of the disqualifiers that sink strong teams. It is not a probability of winning. Anyone selling you a "guaranteed win" tool is selling theater. This removes every reason to lose that is not the merit of your work, and sharpens the parts that are. That is the whole game. The teams that do this win at a rate the teams that wing it do not.

Paste your draft into the free Reviewer Scorecard. No signup. Five minutes. It tells you the one section that loses you the award, before three weeks of work do. That is what Outrider builds: structure for a process that has been opaque far too long, for the teams that deserve to win it. The full playbook starts with the main guide.