An uncompetitive budget usually is not too expensive — it is unexplained.

Most budgets that fail do not fail on total cost. They fail because a reviewer could not see the reasoning, could not reconcile the numbers with the work plan, or found an arithmetic error that made them distrust everything else on the page.

The starting discipline is derivation. A budget should be built from the work plan, activity by activity, rather than assembled to hit a target and reconciled afterward. Reviewers check the correspondence, and a budget containing a $30,000 line that appears nowhere in the narrative — or an activity in the narrative with no cost attached — reads as two documents written by two people who did not speak.

Personnel is the largest line in most awards and receives the most scrutiny. State each role's function in the project rather than its title, and justify the level of effort with something other than assertion. A ratio that produced a documented outcome is the strongest available justification: two coordinators because the caseload ratio behind an 84 percent retention rate was one to thirty-five, and this cohort is seventy. Partial FTEs need an explanation of who pays for the rest, or they read as budget-filling. Fringe should be stated as a rate with a basis, not a lump.

Indirect costs are where the most avoidable value is lost. Organizations routinely request nothing, or a token amount, believing it improves their odds. It does not; it signals inexperience and it starves the finance and compliance functions the funder itself depends on. If you have a negotiated rate, use it and say so. If you do not, the de minimis rate under the 2024 Uniform Guidance revision is 15 percent of modified total direct costs, and you are entitled to elect it. If the funder caps below your rate, state the cap, apply it, and note in one sentence what the organization is absorbing. Silent absorption teaches every funder that your true costs are lower than they are.

Allowability deserves a deliberate pass rather than an assumption. Organizations under-claim constantly — allowable costs left off include participant support, evaluation, staff training tied to the project, translation and interpretation, accessibility accommodations, data systems, audit costs, and equipment within thresholds. Meanwhile a small number of unallowable items appear repeatedly: fundraising, lobbying, entertainment, most alcohol, and bad debt. Reading the cost principles once for your funder type is a couple of hours that pays for itself immediately.

Match and cost share must be treated as a commitment, not a rounding gesture. Know whether the requirement is cash or in-kind, whether federal funds may count (usually not), and how it will be documented at closeout. Inflated in-kind figures are transparent to experienced reviewers and create a real compliance obligation if you win. A smaller, committed, documentable match is worth more than a large speculative one.

Then there is the arithmetic, which is the most preventable failure of all. Totals that do not sum, percentages applied to the wrong base, a fringe rate inconsistent between the table and the narrative, a year-two escalation applied to one line and not another. Every one of these is fatal to reviewer confidence out of proportion to its size, because a reviewer who finds one error assumes there are others. The rule is simple: the budget is checked by someone who did not build it, before submission, every time.

Multi-year budgets carry their own traps. State escalation assumptions explicitly. Show start-up costs in year one only and explain why they do not recur. If the funder expects a sustainability ramp, show the ramp in the numbers rather than promising it in prose.

The narrative alongside the budget should justify shape, not restate the table. The reviewer already has the table. What they need is why this many people, why this rate, why this proportion to overhead, and what the number is sensitive to. Including a short sensitivity paragraph — what gets cut first at 70 percent of the request, what breaks below 50, which line is genuinely fixed — does two things: it reads as operational maturity, and it frequently prevents an arbitrary trim, because the reviewer now knows that cutting the evaluation line does not save eight percent, it invalidates the outcome data they want in the final report.

A note on right-sizing. Applicants sometimes reduce a request believing a smaller ask is easier to fund. Against a program with a published range, an unusually low request suggests you have misunderstood the scope. Ask for what the work costs, justify it, and let the funder negotiate.

The final test is external and takes ten minutes. Hand the budget and narrative to someone who has never seen your programs and ask them to explain, in their own words, why you need this much money. If they can, a reviewer can defend it in committee — which is the actual mechanism by which budgets survive.