An SBIR award is meant to build capability inside a small business, so the program limits how much of the work you may hand to anyone else. The company must perform a minimum share of the research itself, that minimum is higher in Phase I than in Phase II, and STTR replaces the whole arrangement with a different split. Getting this wrong is not a scoring penalty; it can make a proposal ineligible.
How the limit works
The rule is expressed as a minimum percentage of the research or analytical effort that the small business must perform, generally measured in budget dollars rather than in hours. Everything performed by anyone outside the company counts against the remainder: university subawards, contract research organizations, testing labs, and individual consultants alike. Founders often assume consultants are exempt because they are people rather than institutions. They are not.
Phase I allows a smaller share of outside work than Phase II, on the logic that a feasibility study should be proving that your team can do this. Phase II loosens the constraint because building and validating a real system reasonably involves specialists. The specific figures are set in statute and restated in SBA's SBIR Policy Directive and in each agency's solicitation, and agencies may grant a waiver in particular circumstances, so take the current numbers from the solicitation you are answering rather than from an article. The phase difference is one more reason to understand what separates a Phase I feasibility study from a Phase II build before you shape the team.
STTR has its own split
STTR exists precisely for the project that leans on an outside research partner. Instead of a ceiling on outsourcing, it prescribes a floor for both sides: the small business must perform at least a set share, a single partnering research institution must perform at least its own set share, and the remainder may go to either. The partner must be a research institution of the type the program defines, the relationship needs a written agreement covering IP allocation, and the principal investigator may sit at either organization — a flexibility SBIR does not offer.
If your plan hands the majority of the science to a university lab, you are usually looking at an STTR rather than a stretched SBIR. That choice, along with everything else that separates SBIR from STTR, should be made before you build the budget, because retrofitting one into the other rarely works cleanly.
Structuring a partner-heavy project
When an outside group is central to the work, the practical move is to draw the task boundary so that the company keeps the parts that are genuinely yours. Keep system integration, design decisions, test planning, data analysis, and the build of anything that becomes your product in-house, and buy the discrete specialist service — a characterization run, an assay, a certification test — from outside. That usually satisfies the limit without pretending.
Then check the arithmetic in your budget rather than trusting the narrative, since the percentage is computed from dollars. Subaward and consultant lines are also where budget mistakes that sink applications cluster. And if the partner relationship is going to outlive the award, negotiate the IP and publication terms early, the way any serious research partnership with a university lab should be set up.
This is general information, not legal or accounting advice — confirm the current limits with the solicitation and your own advisors.
Projects House performs the in-house engineering share of federally funded development for small companies. Talk to us through the contact form.