How a Bad Budget Loses a Federal Grant

Reviewers rarely reject a proposal because the science is wrong. They reject it because the budget tells a different story than the work plan. A budget that shows one engineer part-time against a schedule promising three prototypes, or a $40,000 line for equipment in a program that does not allow capital purchases, tells the panel that the applicant has not actually planned the project. The technical narrative gets the enthusiasm; the budget gets the skepticism.

Federal award budgets follow one rulebook — the uniform administrative requirements and cost principles covering nearly all federal grants. Learn the vocabulary once and it applies at NIH, NSF, DOE, DoD and most state pass-through programs. Here are the mistakes that cost applicants awards.

Mistake 1: not knowing what is direct and what is indirect

A direct cost can be identified with one specific project: the engineer building your prototype, the parts in it, the test lab invoice. An indirect cost benefits the whole organization and cannot be traced to one project without unreasonable effort: rent, utilities, accounting, general administration, the office manager.

The error is moving an indirect-type cost into the direct column because it looks more concrete there. Office supplies, general software licenses, telephone service and administrative salaries are normally indirect. Billing them directly on one award and indirectly on another creates an inconsistent treatment problem, which is exactly what auditors look for. Pick a treatment and apply it everywhere.

Mistake 2: guessing at the indirect rate

You recover indirect costs one of two ways. Either you negotiate a rate with your cognizant federal agency, producing a negotiated indirect cost rate agreement that says what percentage you may charge and what base it applies to, or you use the de minimis rate available to organizations that have never had a negotiated rate. The de minimis is applied to modified total direct costs — a defined base that excludes equipment, capital expenditures, participant support, rent and the portion of each subaward above the first $25,000.

Three things go wrong here:

• Applying the rate to the wrong base. Charging your rate against total project cost instead of modified total direct costs inflates the request and gets corrected downward at award, leaving you short.
• Assuming a program allows full indirect recovery. Many do not. SBIR solicitations, state programs and foundations frequently cap indirect at a fixed percentage or disallow it entirely.
• Forgetting that an indirect rate agreement, once you have one, is binding. You cannot quietly use the de minimis instead because it is higher.

Small companies routinely under-recover here, then discover mid-project that overhead is eating the founder's salary. If you are weighing whether the paperwork is worth it at all, that calculation belongs in the same conversation as grants versus equity funding.

Mistake 3: fringe benefits invented on the spot

Fringe is payroll taxes, health insurance, retirement contributions and paid leave, expressed as a percentage of salary. Applicants either omit it, which understates true cost, or pick a round number with nothing behind it. Compute it from actual payroll: total benefit cost divided by total salary cost. A defensible low number beats an undefended average.

Related error: budgeting a founder's salary at a rate the company has never paid. Federal awards reimburse actual compensation, supported by time records.

Mistake 4: unallowable costs sitting in plain sight

Some costs are never reimbursable on a federal award, no matter how genuinely the project incurred them. The ones that show up most often in first-time budgets:

Patent costs deserve a note because founders assume grant money can pay for filings. Usually it cannot, even though the award may create obligations about who owns the resulting invention. Read who owns the IP from a federal grant before you plan around it.

Mistake 5: equipment and supplies in the wrong buckets

Equipment has a federal definition: tangible property with a useful life over one year and a cost at or above your capitalization threshold, commonly $5,000. Below that it is supplies. The distinction matters:

• Equipment is excluded from the modified total direct cost base, so you earn no indirect recovery on it.
• Equipment carries property management, inventory and disposition rules that follow you to closeout.
• Many programs — SBIR Phase I especially — restrict or prohibit equipment purchases outright.

Splitting one $12,000 instrument into three purchase orders to keep it under the threshold is not a workaround; it is exactly the pattern an auditor is trained to find. If the program will not buy the machine, budget for outside services instead. Sending work to a vendor is often cheaper anyway, as the comparison in buying a printer versus using a service shows for one common case.

Mistake 6: subawards and consultants treated the same way

A subaward transfers part of the scope of work to another organization performing a substantive portion of the project. A contractor or consultant supplies goods and services within normal business operations. Programs treat them differently, and several cap how much work may be performed outside the small business, with different limits by phase. Exceeding the cap is an eligibility failure, not a scoring deduction.

Practical rules: name the subrecipient and attach their scope, budget and rate agreement; apply the $25,000 rule to your indirect base; use documented consultant day rates. A university lab will apply its full negotiated rate to its portion — worth knowing before you commit, as in partnering with universities on product R&D.

Mistake 7: cost share you cannot actually document

If a program requires matching funds, or you volunteer them, everything you promise becomes auditable. Donated time needs timesheets at documented rates; donated equipment needs a defensible valuation. Another federal award cannot serve as match, and investor money that has not closed is not a commitment. Unmet cost share at closeout can reduce what you keep.

Voluntary cost share rarely improves your score and always increases your compliance burden. Offer it only when the solicitation says it matters.

Mistake 8: a budget justification that contradicts the work plan

This is the one that most reliably lowers a score. The narrative promises testing at three sites; travel funds one trip. The schedule shows a firmware engineer for nine months; the personnel table shows 0.2 full-time equivalent. The commercialization section describes regulatory testing; nothing pays for it.

Fix it by writing the justification against the task list, task by task. Every task needs identifiable hours, materials and services behind it, and every budget line should trace to a task. Panels read the two documents side by side — as described in how federal review panels score proposals.

Build the budget first, not last

The budget is the most reliable statement of what you intend to do. Build it early, from a real task breakdown with real quotes for parts, tooling, test lab time and certification, then write the narrative to match. Applicants who work in that order write tighter proposals, because the cost of each ambition becomes visible while there is still time to cut scope. It also produces honest numbers if you raise privately instead — the discipline behind financial projections investors trust.

Projects House builds the engineering plans, task breakdowns and cost estimates that grant budgets rest on — real quotes rather than placeholders. If you are preparing an application and need the numbers to hold up, get in touch through our contact form.