TL;DR
Grant management is the administration of awarded funding against the funder’s rules: what may be charged, how it must be documented, who must be monitored, and what must be reported. Its defining difficulty is that a cost can be properly approved, correctly coded and within budget, and still be unallowable, because allowability is determined by the award terms rather than by your own approval policy.
Key Takeaways: Federal awards are governed by 2 CFR 200, the Uniform Guidance. Every cost must be allowable, allocable, reasonable and consistently treated. The Single Audit threshold rose from $750,000 to $1,000,000 in federal expenditures. Auditors must report known questioned costs above $25,000. The most common findings concern documentation rather than the expenditure itself.
What is grant management?
Grant management is the administration of awarded funding across its full lifecycle: budgeting against the award, determining what may be charged to it, documenting those charges, monitoring any organizations the funds pass through to, reporting to the funder, and closing out.
It applies anywhere federal awards are expended, which is a broader population than the term “grants” suggests. Hospital systems and academic medical centers, universities and research institutions, state and local government, and nonprofits of any size all operate under the same framework.
That framework is 2 CFR Part 200, the Uniform Guidance. It consolidates the rules across every federal agency and every award type, and it overrides individual agency rules where they conflict.
The four tests every cost must pass
A charge to a federal award must satisfy all four, not any one.
- Allowable. Permitted under 2 CFR 200 Subpart E and under the specific terms of the award.
- Allocable. The award benefited from it, in proportion to what was charged.
- Reasonable. What a prudent person would have paid in the circumstances.
- Consistently treated. Handled the same way as the same type of cost on non-federal work. A cost treated as indirect elsewhere cannot be charged as direct here.
That last test is the one organizations most often fail without realizing, because it is not a judgment about the transaction at all. It is a judgment about whether the transaction is consistent with how you treat everything else.
The distinction that defines the discipline
Here is the point worth building a process around.
Your approval workflow tests your rules. The award tests different rules.
A purchase can clear every internal control you operate. The requester had authority. The budget had room. The coding matched the chart of accounts. The approver signed. The invoice was paid correctly.
And it can still be unallowable, because the award says so and your workflow never asked the award.
This is why grant compliance is not a stricter version of ordinary spend control. It is a second, parallel rule set applied to the same transactions, held in award documents rather than in policy, and differing from award to award.
An organization with excellent internal controls and no mechanism for testing transactions against award terms has a well-controlled process that produces findings.
The findings are about evidence
The second structural point follows from the first, and it is visible in how auditors phrase what they find.
A typical allowable-costs finding does not say the money was spent improperly. It says the transaction support does not show purpose, allocability, or award fit.
That is a documentation finding, not a spending finding. The cost may well have been entirely proper. What could not be produced was evidence tying it to the award’s purpose, demonstrating the proportion charged, and showing it fit the award’s terms.
The same pattern appears in the other recurring findings. Subrecipient monitoring findings cite the absence of a risk assessment, a monitoring record, or follow-up on findings, rather than a subrecipient doing something wrong. Time and effort findings cite documentation that does not substantiate how effort was distributed.
The practical implication is uncomfortable. Most of what is at risk in a Single Audit is not the quality of your decisions. It is whether you can evidence them two years later.
What changed, and what is pending
The 2024 revision to the Uniform Guidance moved several thresholds that remain in effect.
- Single Audit threshold: raised from $750,000 to $1,000,000 in federal awards expended in a fiscal year (2 CFR 200.501), effective for fiscal years beginning on or after October 1, 2024.
- De minimis indirect cost rate: raised from 10 percent to 15 percent of modified total direct costs for recipients without a negotiated rate (2 CFR 200.414(f)).
- Equipment capitalization threshold: raised from $5,000 to $10,000.
- Questioned costs: auditors must report known questioned costs above $25,000 as an audit finding (2 CFR 200.516).
Separately, OMB released proposed revisions in May 2026 covering conflicts of interest, mandatory disclosures, award monitoring, allowable costs and subaward management, carrying a proposed effective date of October 1, 2026. That date has now passed without a final rule, and the existing requirements remain in force, including the $1,000,000 threshold. Verify current status before relying on any summary, including this one.
One consequence of the threshold change deserves attention. Organizations that remain above it are reporting rising audit costs, as auditors price in expanded documentation expectations around internal controls and subrecipient monitoring. Crossing out of the Single Audit population does not remove the underlying obligations either: cost principles, financial management standards, reporting, records retention and pass-through monitoring all continue to apply.
Where the work actually goes
Look at what compliance requires transaction by transaction and the shape is clear.
Someone determines, for each charge, which award it belongs to and whether the award permits it. Someone establishes the allocation basis where a cost benefits more than one project, and documents that basis. Someone assembles the support demonstrating purpose, allocability and award fit, which usually means retrieving an invoice, a requisition, an approval, and the award clause it sits under. Someone tracks period-of-performance boundaries, because a cost incurred outside the period is unallowable regardless of merit. Someone maintains subrecipient files: risk assessments, monitoring records, single audit verification, follow-up on findings. And someone reconciles the reported figures back to the general ledger.
Almost none of this is accounting judgment. It is reading award terms, reading transaction documentation, and establishing that the two correspond, at a volume that scales with transactions multiplied by awards.
Which is why it is triaged. Large charges receive full documentation. The long tail is coded and paid, and the evidence is assembled later, if at all, usually under audit, usually by someone reconstructing it from memory and email.
Where automation fits
The constraint is reading award terms and matching transactions against them, not calculating anything, and that is where automation changes the position.
Automation that can read unstructured documents and reason about their contents can interpret award terms and the applicable cost principles, test each charge against the award it is coded to rather than only against internal policy, assemble the support that demonstrates purpose, allocability and award fit at the time of the transaction, flag period-of-performance boundaries and consistency conflicts, and maintain the subrecipient monitoring file as the evidence arrives rather than before an audit.
The shift is from reconstructing evidence to capturing it. A charge documented when it occurs costs a few seconds. The same charge documented two years later, by someone who was not involved, costs considerably more and frequently cannot be done at all.
Because this material becomes audit evidence submitted to a federal cognizant agency, every determination needs to be traceable to the award clause and the document it rests on. A conclusion nobody can substantiate is precisely what a questioned-cost finding describes.
To be clear about scope, Kognitos is not a grants management system. It does not replace the platforms that hold award records, manage budgets and produce funder reports. What it addresses is the layer beneath: reading award terms and transaction documentation, testing one against the other, and producing an evidenced position on each charge.
For related material, see our guides on internal controls, PBC list, regulatory reporting, spend management, and quality management systems. To see how deterministic AI tests transactions against written rules with a full audit trail, book a demo or try the platform.
Getting started
Two exercises, and the second is the revealing one.
Take twenty charges from a single award and try to produce the support an auditor would ask for: purpose, allocability, award fit, and the clause permitting the charge. Time it. The proportion you cannot produce within an hour is your actual documentation position, independent of whether any cost was improper.
Then check consistency rather than allowability. Pick one cost type charged as direct to a federal award and establish how the same cost type is treated on non-federal work. Inconsistent treatment is a finding even where every individual charge was reasonable, and it is the test organizations are least likely to have examined.
