Which DoD components answer your questions, how long do they take, and what should you ask before submitting an SBIR proposal?
DSIP is the Defense SBIR/STTR Innovation Portal. In RallyProp's October 2, 2026 snapshot of 1,841 questions across 11 DoD components, 71% of 1,764 classified answers added information that could change a proposal. Ask specific questions early, but tailor them to the component.
01 / The agency effect
DHA had the highest reported usefulness rate, at 91%, but only 25 questions. Among components with at least 30 questions, MDA led at 79% across 34 questions. Navy combined the largest volume with a 73% usefulness rate across 700 questions. These are historical observations, not forecasts of answer quality or award chances.
Useful means the answer added information that could change a proposal. It does not mean a higher chance of award.
Amber bars: fewer than 30 questions. Percentages are reported in the source, not recalculated from question counts; the denominator is classified answers.
Among the large components with observable timing, Air Force / Space Force was fastest at a 2.9-day median, followed by Navy at 4.1 days and OSD / SCO at 14.1 days.
Median observed time from first question sighting to first answer sighting, not an agency service commitment. Army, SOCOM, MDA, and DTRA timing is not observable.
75%
Pre-open
69%
Open period
Yes: in this sample, 75% of pre-open answers were useful, compared with 69% during the open period. Ask early and specifically; the guide finds agency differences more consequential than the timing window.
02 / Proposal decisions
Ask quantified clarification questions, check for changes to requirements, and prepare a proposal that does not depend on receiving an answer.
Quantified, either/or questions get clearer answers. Ask whether 50 W means average or peak; avoid asking whether your solution is a fit.
Answers corrected budgets, page limits, units, and even topic objectives. Re-read the board before locking the technical and cost volumes.
Data and equipment are often unavailable in Phase I. Define defensible metrics, identify your own datasets, and confirm clearance gates early.
03 / Field intelligence
Match the question to the component's answer patterns: ask Navy about scope and transition, DAF about thresholds, Army about interfaces, and OSD / SCO about clearance gates. The original downloadable guide includes the full topic-level examples, deflection patterns, and proposal recommendations.
Ask quantified, either/or questions. Ask early on MBSE and AEGIS topics; late questions can sit for weeks.
700 questions / 4.1d median / 90th percentile: 18 days
Data access, transition partners, scope boundaries
Q&A shifted the IETM objective from a conversion tool to a single agnostic viewer.DON26BZ05-NV078
Watch for: Phase I data and equipment are usually not provided. Open topics often redirect to industry-day PDFs.
Separate thresholds from objectives. Read AFWERX and SpaceWERX FAQs before asking.
523 questions / 2.9d median / 90% within 8 days
Thresholds, SWaP, CONOPS, D2P2 feasibility rules
Answers specified 50 W orbit-average power and a 15 kg ceiling for the whole awareness payload.DAF26BX06-DV511
Watch for: Q&A cannot pre-judge D2P2 feasibility. Plan your own Phase I datasets.
Ask before open and frame questions as specification clarifications, not a pitch for your approach.
184 questions / Timing not observable / Only answered questions visible
Scope edges, interfaces, specs, Phase I deliverables
The 2 lb limit applied to compute only; passive collaboration meant no inter-UAS communications.ARM26BX04-NV008
Watch for: Public timestamps cannot reveal response time. Approach feedback and one-on-one meetings are usually declined.
Confirm FCL requirements early. Define and justify your own metrics when the topic leaves them open.
168 questions / 14.1d median / SCO may answer in late batches
Clearance gates, period of performance, data
Q&A clarified a 12-month D2P2 period of performance despite longer durations in topic text.OSW26BZ05-DV022
Watch for: Do not make the proposal schedule depend on an answer. The observed median was two weeks.
Request the NDA or data-use agreement as soon as the topic posts. Show a proven transition-ready capability.
94 questions / ~1d median / Approximate; only 5 timed answers
Baselines, thresholds, NDA and data access
DLA named a 10% revenue threshold and parent-level gross revenue as the denominator.DLA26BZ03-NV012
Watch for: One topic accounts for 66 of 94 questions. Much detail is NDA-gated and some public answers are empty.
Compare the Statement of Objectives with the DSIP text and ask about each conflict.
68 questions / Timing not observable / Only answered questions visible
Mission scenario, users, hardware envelope
The UUV LiDAR envelope was a 5-inch OD, at most 1 m long, rated to 200 m.SOC26BZ01-DV001
Watch for: Treat passive-only and envelope limits as gates. Format questions usually get redirected to instructions.
Ask whether a technology is in or out. Derive performance metrics from a defensible CONOPS.
34 questions / Timing not observable / Only answered questions visible
Hard exclusions, bands, test scope
Shielding, including 3D-printed shields, was explicitly outside the rad-hardening topic.MDA26BZ04-NV002
Watch for: Small sample. Numeric targets may be withheld because platforms are classified.
Ask precise spec questions; this sample contained short, decisive answers.
25 questions / 7d median / Small sample
Size, power, specs, Phase I scope
A particulate sensor answer clarified a limit of 20 cubic inches, not a generic size target.
Watch for: The 91% rate is from only 25 questions and is not a forecast.
Ask direct scope questions and verify which release controls the deadline.
23 questions / 2.8d median / Small sample
In-scope confirmations, release-specific eligibility
Q&A confirmed a hybrid stationary reader plus far-forward dipsticks was in scope.
Watch for: Administrative and STTR eligibility questions often repeat the instructions.
Clarify required versus desired bands and check foundry and license constraints.
15 questions / 5.9d median / Small sample
Frequency bands, foundry constraints
A printed RF MCM answer separated the required 100-450 MHz band from the desired 100 MHz-1 GHz range.
Watch for: Only 15 questions across 3 topics; do not generalize the technical constraints across DMEA.
Write to the topic text and keep moving. The sampled CWMD topic had no technical POC assigned.
7 questions / Timing not observable / No observable response time
Do not depend on a technical answer
Every sampled CWMD question received the same letter without technical clarification.
Watch for: Seven questions on two topics, not an agency-wide policy or a prediction.
04 / The complete research guide
All agency profiles, topic-level examples, recurring non-answers, and proposal tips. Download the original October 2 research guide as a PDF. This public page is the question-led AEO edition; the PDF retains the original research presentation.
Free download. No account required. Weekly research updates are optional.
DSIP Q&A is the question-and-answer board for topics on the Defense SBIR/STTR Innovation Portal. Applicants use it to clarify technical requirements, scope, and proposal instructions. In RallyProp's October 2, 2026 snapshot, 71% of 1,764 classified answers added information that could change a proposal.
Often not. In the sampled Navy and SpaceWERX answers, applicants were usually expected to use synthetic or public data in Phase I. This is not a universal rule: a SOCOM UUV LiDAR answer specified two government DREWUV frames. Confirm data and equipment availability in the specific topic's official Q&A before planning your work.
A recurring answer in the 2026 sample placed CMMC Level 2 (Self) at award, not proposal submission. That observation is not a universal compliance rule. Check the current solicitation, component instructions, and official Q&A for your topic's assessment level and deadline.
No prospective feasibility determination was offered in the sampled SpaceWERX D2P2 answers. Applicants should document feasibility against the solicitation's requirements rather than rely on Q&A to pre-approve a prototype or approach.
The guide reports 39 active DARPA topics in 2026 but no DSIP Q&A for them. Use DARPA's own topic-specific question channels and instructions instead of waiting for answers on the DSIP board.
Yes. The sampled answers corrected budgets, page limits, units, and technical scope. For Navy topic DON26BZ05-NV078, Q&A shifted the IETM objective from a conversion tool to a single agnostic viewer. Re-read the current official topic and Q&A before finalizing your technical and cost volumes.
RallyProp analyzed Q&A change snapshots joined to topic components and open/close dates. This is an FY-2026 snapshot published October 2, not a full-year census. RallyProp observed 1,841 questions first seen in 2026, of which 1,765 were answered. The usefulness headline uses 1,764 classified answers.
Labels are one reviewer's judgment. Timing comes from change snapshots collected about hourly on weekdays; answers posted with questions have no observable response time. Small samples and topic concentration limit agency-wide conclusions.
Topic examples are historical, topic-specific observations, not current eligibility, clearance, or compliance advice. Confirm the latest solicitation and official Q&A before making a proposal decision.
Explore DSIP opportunities