Writing a Cost Allocation Plan With AI: Multi-Program Nonprofits
The moment an organization runs a second program funded by a second grant, a question appears that never had to be answered before: when the executive director spends a morning on something that serves both, who pays for the morning? A cost allocation plan is the written answer. It is a real document, maintained by the finance lead, explaining how every shared dollar is divided and why that division reflects the benefit each program actually received. Auditors read it. Funders ask for it. And organizations that have never written one down discover, usually at the worst possible moment, that a practice living in one person's spreadsheet is not the same thing as a methodology.
A cost allocation plan is a written description of how your organization assigns costs that benefit more than one program, grant, or cost objective. It names the cost pools, states the base used to split each pool, explains why that base reflects relative benefit, identifies the data source behind the base, and says how often the numbers get recalculated. It is not a spreadsheet of percentages. The percentages are the output. The plan is the reasoning that produces them, written down in a form that someone who has never met you can follow.
The requirement that sits behind all of this is 2 CFR 200.405, which says a cost is allocable to a federal award if it is assignable to that award in accordance with the relative benefits received. When a cost benefits two or more projects in proportions that can be determined without undue effort, it must be allocated on that proportional benefit. When the proportions genuinely cannot be untangled because the work is interrelated, the costs may be allocated on any reasonable documented basis. Note what that second clause requires. Not a defensible instinct. A documented basis.
Whether you are formally required to produce a standalone plan document depends on your funders. Some federal awarding agencies and pass-through entities request one explicitly, particularly when an organization does not have a negotiated indirect cost rate. Some state agencies and large private funders ask for the methodology as an attachment to the budget narrative. Many nonprofits are never asked directly and still need one, because the alternative is being unable to explain a number under questioning, which is functionally the same as the number being wrong.
This article covers the plan as a document you write, defend, and maintain: what belongs in it, how to choose and justify allocation bases, how to document payroll so it survives 2 CFR 200.430 scrutiny, what to do about staff split across three or more grants, how to true up when programs change, and how to keep the plan consistent with your audited financials and your Form 990. It also marks where AI genuinely reduces the work, which is more places than most finance leads expect, and where it has no business at all.
Three Adjacent Documents That Get Confused
Before going further it is worth separating three things that live near each other and are routinely treated as one. Confusing them is the single most common reason a finance conversation goes in circles, and it is also why organizations sometimes solve one problem while believing they have solved all three.
The statement of functional expenses splits your total expenses three ways for external reporting: program services, management and general, and fundraising. It is a presentation requirement, it appears in your audited financials and on Form 990 Part IX, and its audience is donors, watchdogs, and the public. Our guide to defensible program, admin, and fundraising splits covers that document and the judgment calls inside it.
An indirect cost rate is a single negotiated or elected percentage that lets you recover overhead on federal awards without itemizing it. You either negotiate a rate with your cognizant agency or elect the de minimis rate, which the 2024 revisions to the Uniform Guidance raised from 10 percent to 15 percent of modified total direct costs. Our guide to negotiating an indirect cost rate deals with the proposal, the base, and the negotiation itself.
A cost allocation plan, the subject of this article, is the internal methodology that divides shared and direct costs horizontally across your programs and your grants. It answers a different question from either of the others: not how much of the total is program versus admin, and not what percentage you may charge for overhead, but how the rent, the shared case manager, the database subscription, and the fiscal officer's time get split among Program A, Program B, the state contract, and the family foundation grant.
The three are related but not substitutes. A well-built cost allocation plan feeds the functional expense statement, because the program-level allocations roll up into the program services column. It also feeds an indirect cost rate proposal, because the plan is where you demonstrate which costs are being treated as indirect and how the remainder is distributed. Organizations that build the allocation plan first usually find the other two considerably easier. Organizations that skip it end up reverse-engineering a methodology from numbers they already reported, which is a bad position to argue from.
Which document answers which question
Pick the right one before you start building
- How much of our spending is program versus admin versus fundraising? The statement of functional expenses
- What percentage may we charge a federal award for overhead? The indirect cost rate
- How do we divide this shared cost across four programs and six grants? The cost allocation plan
- Why did Program B absorb 34 percent of rent this year and 28 percent last year? The cost allocation plan, again
Direct, Shared, and Indirect: Where the Lines Fall
The plan begins by sorting every cost the organization incurs into one of three treatments. A direct cost is identifiable specifically with a single final cost objective, which in practice means one program or one award. The salary of a caseworker who works only on the workforce program is direct to that program. A shared cost, sometimes called a joint or common cost, benefits two or more cost objectives in proportions that can be reasonably estimated. Rent for a building housing three programs is shared. An indirect cost is incurred for common objectives and cannot be readily assigned to a specific final cost objective without effort disproportionate to the result, which is where general administration typically sits.
The line between these categories matters more in a multi-program organization than in a single-program one, for a reason that becomes obvious once you see it. In a single-program nonprofit, a cost misclassified as indirect when it was really direct mostly affects presentation. In a multi-program nonprofit with multiple grants, that same misclassification moves real dollars between funders. A cost pushed into the indirect pool gets recovered through your rate, if you have one, spread across all awards according to their base. The same cost treated as direct lands entirely on one grant. Two different funders end up paying two different amounts because of a classification decision, which is exactly why auditors examine the decision.
This is the force behind the consistency requirement in 2 CFR 200.403. Costs must be accorded consistent treatment, and specifically a cost must not be charged to a federal award as a direct cost if any other cost incurred for the same purpose in like circumstances has been allocated to that award as an indirect cost. You cannot treat the bookkeeper's time as indirect when working on the federal grant and direct when working on the state contract because the state contract happens to allow it. Whichever way you go, you go that way everywhere, and the plan is where you commit to it in writing.
A practical way to test your classifications is to take each cost and ask whether the organization would still incur it if a particular program disappeared tomorrow. Costs that would vanish entirely are direct to that program. Costs that would shrink proportionally are shared and need a base. Costs that would continue at nearly full strength regardless are indirect. The test is not perfect, and it will not resolve every borderline item, but it exposes the cases where a cost has been classified by habit rather than by reasoning. Those are the ones to write down and defend explicitly.
The infrastructure that makes all of this workable is the coding structure underneath. A chart of accounts built with grant and program dimensions lets you capture direct costs at the point of entry rather than sorting them out later. Organizations that code a single natural account for salaries with no program or funder dimension are guaranteeing that every allocation is a reconstruction, and reconstructions do not survive review well.
Choosing an Allocation Base and Matching It to the Pool
The heart of the plan is the pairing of each cost pool with a base that measures relative benefit for that specific pool. The Uniform Guidance does not prescribe bases. It requires that whatever base you use produce results reflecting relative benefit, be documented, and be applied consistently. That freedom is a trap for organizations that pick one base and apply it to everything, because a base that is reasonable for rent is frequently unreasonable for insurance, and the mismatch is visible to anyone who thinks about it for thirty seconds.
Square footage is the natural base for occupancy costs: rent, utilities, janitorial, building insurance, property taxes where applicable. It works because the benefit each program derives from the building really does track the space it occupies. The complications are common areas and unassigned space. A defensible approach allocates dedicated program space directly, then distributes common areas such as hallways, restrooms, the kitchen, and the conference room in proportion to the dedicated space, so common area does not silently become a subsidy for whichever program the finance lead favors. Space that sits genuinely vacant should not be charged to programs at all.
Full-time equivalents work well for costs that track headcount rather than activity: general liability insurance in some structures, human resources support, payroll processing, per-seat software licenses, and staff training budgets. FTE is easy to compute and easy to verify, which is part of its appeal. Its weakness is that it treats a part-time intake worker and a program director as equivalent consumers of whatever is being allocated, which is fine for a per-seat license and much less fine for anything correlated with salary level.
Direct labor dollars or modified total direct costs suit administrative and financial support functions, because the work performed by finance, HR, and executive leadership genuinely does scale with the dollar volume of activity being supported. A program spending $900,000 consumes more of the bookkeeper's attention than one spending $90,000. This is also the base most compatible with an indirect cost rate proposal, which is worth knowing if you expect to pursue one later. Its main hazard is that programs dominated by subawards or direct participant assistance carry large dollars with little administrative burden, which argues for a modified base that excludes those elements.
Units of service is the strongest base when you have reliable counts and the cost genuinely varies with volume: meals served, sessions delivered, clients enrolled, bed nights provided. It is also the base most likely to be contested, because it depends on program data that may be collected inconsistently across programs. If two programs count a client in materially different ways, the base is not comparable and the allocation is not defensible. Use it where your data is solid and do not use it where it is aspirational.
Headcount served, distinct from FTE, allocates certain program support costs by the number of individuals reached. It is intuitive to funders and easy to explain, which makes it attractive in budget narratives, but it can badly misstate benefit when programs serve populations with very different intensities. A drop-in service reaching eight hundred people briefly and an intensive case management program serving forty people over a year are not comparable on headcount, and pretending otherwise is the kind of thing a reviewer notices immediately.
The mistake worth naming explicitly is allocating by revenue. It is tempting because the data is easy and the result looks proportional, but it inverts the logic entirely: the base is supposed to measure benefit received, not funding available. Allocating shared costs by grant size means a newly awarded grant instantly absorbs a larger share of costs that did not change at all, and it produces the appearance of charging each funder in proportion to its willingness to pay. Auditors treat revenue-based allocation as a red flag, and they are right to.
Sensible pool and base pairings
Each pool gets the base that measures its benefit
- Rent, utilities, janitorial: square footage, with common areas prorated
- HR, payroll processing, per-seat licenses: full-time equivalents
- Finance, executive oversight, audit fees: direct labor dollars or modified total direct costs
- Program supplies and materials: units of service, where counts are reliable
- Shared staff time: actual effort records, not a fixed budgeted percentage
Bases that draw questions
Defensible only in narrow circumstances, if at all
- Grant revenue or award size, which measures funding rather than benefit
- Equal splits across programs of obviously unequal size
- Percentages carried forward from a budget written three years ago
- Whatever remaining capacity a grant happens to have at year end
- A single base applied uniformly to every cost pool in the organization
Drafting the Methodology Narrative
The written plan is a short document, usually somewhere between five and fifteen pages depending on how many programs and pools you have. Length is not the point. The point is that a reader who knows nothing about your organization can follow the reasoning from a dollar entering your general ledger to a dollar charged against a specific award, and can reproduce the calculation from the sources you name.
Open with the organizational context: what the organization does, what programs it operates, what funding streams support them, and what the fiscal year is. Reviewers who understand the shape of the organization interpret the rest of the plan more generously, and the section takes half a page. Follow it with the accounting framework, meaning the basis of accounting, the fund or class structure in your accounting system, and how program and funder dimensions are captured at entry.
The core is then a section per cost pool. Each one should name the pool and the accounts it contains, state the allocation base, explain in plain language why that base reflects relative benefit for this particular pool, identify the data source and who maintains it, state the frequency of recalculation, and show a worked example with real numbers from a recent period. The worked example matters more than most people expect. It is the difference between a plan that describes an intention and a plan that demonstrates a practice.
Then handle the exceptions honestly. Every real organization has costs that do not fit the pattern: a grant whose terms prohibit charging a particular category, a program with a funder-mandated allocation method that differs from your standard, a piece of equipment used overwhelmingly by one program but purchased centrally. Write these down as named exceptions with their reasoning rather than letting them sit as unexplained anomalies in the ledger. An exception that is documented is a decision. An exception that is undocumented is an error waiting to be characterized as one.
Close with governance: who owns the plan, who approves it, when it gets reviewed, what triggers an off-cycle revision, and how changes are documented. Most plans should be formally reviewed at least annually and approved by the board finance committee or the board itself. The annual review is not a formality. A recurring single audit issue involving allocation plans is not a bad methodology, it is a good methodology from four years ago that no longer describes what the organization does.
Sections a complete plan contains
Written so a stranger can reproduce your numbers
- Organizational and program overview, with the funding streams named
- Accounting framework and how program and funder codes are captured
- Definitions of direct, shared, and indirect as this organization applies them
- One section per cost pool: accounts, base, rationale, data source, frequency, worked example
- Personnel methodology and the time documentation that supports it
- Named exceptions, with the funder requirement or reasoning behind each
- True-up procedure and what triggers an off-cycle recalculation
- Governance: owner, approver, review cycle, and a dated revision history
Payroll: Making the Allocation Survive 2 CFR 200.430
Personnel is usually the largest allocated pool and it is where the most findings originate, so it gets its own treatment in the plan. The governing standard is 2 CFR 200.430, which requires that charges to federal awards for salaries and wages be based on records that accurately reflect the work performed. The records must be supported by an internal control system providing reasonable assurance that the charges are accurate, allowable, and properly allocated, and they must account for the employee's total activity, not merely the portion charged to federal awards.
That last point is the one organizations most often get wrong. A timesheet showing 60 percent on a federal grant and nothing else does not satisfy the standard, because it does not establish what the other 40 percent was. The record has to encompass both federally assisted and all other activities, so that the federal share is demonstrably a share of something whole. Practically this means every employee whose time touches a federal award needs a record covering 100 percent of their compensated time, including administrative work, fundraising, leave, and time on non-federal programs.
The regulation is deliberately not prescriptive about the form of the record. It does not mandate a particular timesheet template or a specific certification interval. It sets standards the records must meet, and it leaves the implementation to you. What it will not accept is budget estimates alone. Budget estimates may be used for interim accounting purposes, but they must be adjusted so the final charges are based on records reflecting actual work, and any significant variance has to be identified and corrected. Charging a person to a grant at the budgeted 50 percent for twelve straight months, when their actual effort moved substantially, is one of the most reliably cited errors in this area. Our guide to effort reporting on federal grants covers the mechanics of building records that hold up.
Where records do not meet the standard, the federal government may require personnel activity reports with prescribed certifications, which is to say the remedy for weak documentation is heavier documentation imposed from outside. That is worth knowing because it reframes the cost calculation. The work of building a decent effort record now is smaller than the work of operating under a mandated regime later, and considerably smaller than repaying disallowed salary costs.
Staff Who Split Across Three or More Grants
Two-way splits are manageable. The difficulty scales sharply at three and above, because the effort involved in tracking fine-grained time starts to feel disproportionate to the dollars being moved, and because the person doing the splitting is often the same person least able to spare the time. This is where plans quietly degrade into fixed percentages that nobody revisits.
The first structural question to ask is whether the split is necessary at all. Some organizations spread a position across five awards at 15 or 20 percent each when a cleaner arrangement would charge the position directly to the two awards that genuinely drive the work and route the remainder through the indirect pool. Fewer, larger splits are easier to document and easier to defend than many small ones, and a 5 percent allocation to a fourth grant is rarely worth the documentation burden it creates or the audit exposure it opens.
Where a genuine multi-way split exists, the plan should describe how time is captured, at what granularity, and how it is reviewed. Daily or per-activity capture is more accurate than reconstructing a month at the end of it, and the accuracy gap is not small. People reconstructing a month reliably produce numbers close to the budget, because the budget is what they remember. Requiring entries at least weekly, tied to actual activity, breaks that anchoring effect. The supervisor review matters too: certification by someone with firsthand knowledge of the work, rather than the employee alone, is what the internal control expectation is reaching for.
Then watch for the specific failure mode of multi-grant staff, which is that their allocations drift toward whichever grant has remaining budget. A grant running underspent in month ten starts absorbing a little more of the shared coordinator's time, not through any deliberate decision, but because that is where there is room. This is charging based on funding availability rather than work performed, and it is unallowable regardless of how understandable the pressure behind it is. It is also detectable, because the pattern shows up clearly when allocation percentages are plotted against grant burn rates over time.
Grants with restricted purposes add another layer, since the allocation has to respect not just proportional effort but the scope limits in each award. Time a shared staff member spends on an activity outside a particular grant's approved scope cannot be charged there at any percentage. Keeping the restriction tracking on your funds aligned with the allocation methodology prevents the situation where an allocation is arithmetically correct and still charges a cost to an award that does not permit it.
Handling a position split four ways
What the plan should specify for multi-grant staff
- Whether the split is necessary, or whether fewer direct charges plus indirect is cleaner
- Capture interval, with weekly or finer entry rather than month-end reconstruction
- Coverage of 100 percent of compensated time, including leave and non-federal work
- Supervisor certification by someone with firsthand knowledge
- Scope check against each award's allowable activities, not just the percentages
- A periodic comparison of allocation percentages against grant burn rates
Recalculating and Truing Up as Programs Change
A cost allocation plan is a description of a moving organization, and organizations move. A program wins a large new grant and doubles its staff. A contract is not renewed and a program winds down over five months. Two programs merge. A team relocates to a different floor. Each of these changes the correct allocation, and a plan that does not respond to them is describing an organization that no longer exists.
The standard rhythm is interim allocation on established rates with a periodic true-up against actuals. Monthly charges go out on the current percentages, which keeps the books moving and keeps drawdowns reasonable, and then at defined intervals you recompute the base from actual data and adjust. Quarterly true-ups suit most multi-program organizations. Annual-only true-ups are common and are usually too infrequent, because a twelve-month drift produces an adjustment large enough to be disruptive and awkward to explain to a funder whose award has already closed.
Write into the plan the events that force an off-cycle recalculation rather than waiting for the next scheduled one. A new award above a materiality threshold you set, a program ending, a move or significant change in space use, a reorganization that changes reporting lines, and a change in a funder's allowable cost terms all qualify. Naming these in advance converts a judgment call made under pressure into a procedure that simply runs, which is exactly what you want when the finance lead is already busy.
Programs that end deserve particular care because they create a distinctive distortion. When a program closes, its share of shared costs has to be redistributed to the remaining programs, and those costs do not disappear just because the funding did. The rent is the same rent. Organizations that fail to redistribute end up with an unallocated residue sitting in administration, which inflates their overhead ratio and misstates the true cost of the surviving programs. Handling this correctly at the point of closure, alongside the other grant closeout tasks, is considerably easier than reconstructing it during the audit.
Every true-up should leave a record: the period covered, the data pulled, the recomputed percentages, the resulting adjustment by program and award, who approved it, and the date. That record is the single most persuasive evidence available that the plan is a live control rather than a document produced once for a funder and filed. An auditor who sees four dated true-up memoranda for the year asks fewer questions about the methodology, because the memoranda have already answered the question the methodology review was going to raise.
Keeping the Plan Consistent With the Audit and the 990
Your allocation plan does not exist in isolation. The same underlying costs appear in your audited financial statements, in the statement of functional expenses, on Form 990 Part IX, in your grant budgets, and in the financial reports you send each funder. When those presentations disagree, the disagreement is the finding, regardless of which version happens to be correct.
The most common inconsistency is between the plan and the functional expense statement. An organization allocates the executive director's salary 70 percent to programs in its cost allocation plan and reports her as 40 percent program in the functional expense statement, because the two were prepared by different people at different times from different assumptions. Both numbers may be arguable in isolation. Together they demonstrate that no single methodology governs, which is precisely the conclusion you cannot afford an auditor to reach.
The fix is structural rather than cosmetic. One methodology should drive both, with the allocation plan as the source and the functional statement as a rollup of its outputs. If the program-level allocations are correct and the mapping from programs to the functional categories is documented, the functional statement is derived, not separately constructed. The same should hold for Form 990 Part IX, which is why it helps to prepare the 990 from the same allocation data rather than from the prior year's 990 with adjustments. Our guide to preparing the Form 990 covers how that return is read by the people who read it.
Watch the grant budget presentations too. A budget submitted to a funder showing a position at 35 percent, against an allocation plan and effort records showing 22 percent, is not necessarily a problem, since budgets are estimates and actuals differ. It becomes a problem when the charges continue at the budgeted 35 percent all year because that is what the budget said. The plan should state plainly that budgeted percentages are estimates for planning, that charges follow actual effort, and that variances are trued up on the stated schedule.
Where AI Actually Helps With This
Cost allocation suits AI assistance better than most compliance work, for a reason worth stating precisely. The hard parts of the task are writing clear explanatory prose, comparing large sets of numbers for inconsistencies, and anticipating what a reviewer will ask. Those are things language models are genuinely good at. The parts that carry liability, deciding which base is appropriate and standing behind the resulting figures, are not delegated at all, and the boundary between the two is unusually clean here.
Drafting and revising the methodology narrative. Most finance leads know exactly how their organization allocates costs and find writing it down tedious rather than difficult. Describing the pools, bases, and reasoning in conversation and having a model produce a structured first draft, section by section, removes the blank page that keeps these documents unwritten for years. The revision loop is where it earns its keep: tightening a rationale, rewriting a paragraph for a program officer rather than an auditor, or converting the same methodology into a budget narrative attachment.
Sanity-checking allocations against actual activity. Give a model your allocation percentages alongside the underlying activity data, meaning the square footage schedule, the FTE roster, the service counts, and it will tell you quickly where the stated base and the computed base disagree. This is arithmetic comparison across more rows than a person wants to check by hand, and the output is a list of discrepancies for you to examine rather than a set of conclusions to accept.
Spotting drift between the stated base and reality. The highest-value application may be trend analysis. Feed a model twelve or twenty-four months of allocation percentages by program alongside grant burn rates and program activity volumes, and ask it where the percentages move in ways the activity does not explain. Allocation drifting toward underspent grants, a percentage that has not changed while the program tripled, a pool whose split has been frozen since the plan was written: these patterns are visible in the data and invisible to the person producing the entries month by month.
Generating reviewer questions. Hand a model your draft plan and ask it to respond as a single audit senior, then as a federal program officer, then as a skeptical board treasurer, listing every question each would ask and every claim they would want support for. The questions it produces are not exotic, which is the point. They are the obvious ones, and the obvious ones are what you will actually be asked. Answering them in advance and folding the answers into the plan is the cheapest hardening available.
Building the documentation trail. True-up memoranda, revision histories, board approval summaries, and the explanation of why a percentage changed between periods are all short documents that do not get written because they are boring and nobody is waiting for them. Drafting them from the underlying numbers takes a model a minute and is the difference between a plan that is evidently maintained and one that merely exists.
What AI does not do here. It does not set policy. Choosing to allocate occupancy by square footage rather than FTE is a decision about your organization with real consequences for real funders, and it belongs to the finance lead with the board's oversight. A model does not know your lease, your programs' actual space use, or the terms in an award that override your general method. It should not be asked to determine whether a specific cost is allowable, and any regulatory citation it gives you should be opened and read before you rely on it, since confident incorrect citations are the characteristic failure of these tools. Do not upload payroll detail or award documents to a consumer tool without checking the vendor's data handling terms. And the person who signs the plan, presents it to the auditor, and answers for the numbers is a person. That does not move.
Delegate to AI
Checkable against a source you already hold
- First drafts and revisions of the methodology narrative
- Recomputing a base from activity data and comparing it to what you charged
- Trend analysis that surfaces drift between allocations and activity
- Reviewer question lists from an auditor, funder, or board perspective
- True-up memos, revision histories, and approval summaries
Keep with the finance lead
Policy, judgment, and accountability
- Choosing which base applies to which cost pool
- Deciding whether a specific cost is direct, shared, or indirect
- Interpreting an award's terms where they override your standard method
- Approving a true-up adjustment and the entries behind it
- Signing the plan and defending the numbers to an auditor or funder
Defending the Plan in an Audit or Monitoring Visit
If your organization expends $1,000,000 or more in federal awards in a fiscal year, the 2024 revisions to the Uniform Guidance put you in single audit territory, up from the previous $750,000 threshold. Below that you may still face funder monitoring visits, state agency reviews, and pass-through entity site visits, which examine allocation methodology with much the same lens even when the formal apparatus is lighter.
The questions are predictable, which is the good news. An auditor will ask to see the plan, will ask when it was last reviewed and by whom, will pick a month and ask you to walk a specific shared cost from the invoice through the allocation to the charge on a particular award, will ask to see the data behind the base, and will ask whether you applied the same method to non-federal funding. Then they will select a handful of employees and ask for the time records supporting their salary allocations, and they will check whether those records cover the employee's full time or only the federal portion.
The answer that goes badly is the one that begins by explaining the reasoning from memory. Not because the reasoning is wrong, but because reasoning offered verbally in an audit room is treated as an after-the-fact rationalization, which in a documentation sense it is. The answer that goes well is handing over a dated plan, a dated true-up memorandum, and a spreadsheet whose numbers tie to the ledger, and then answering the follow-up questions. The substantive content may be identical. The reception is not.
Assemble the file before anyone asks: the current plan with its approval date, the prior version if it changed during the period, the true-up records, the source data for each base, the time and effort documentation, and the journal entries applying the allocations. Most of this overlaps substantially with what general audit preparation requires anyway, which means the marginal cost of being ready on allocation specifically is small once the broader preparation is in hand.
If you also pass funds through to other organizations, expect questions about their allocation methods as well as your own, since a subrecipient charging you inappropriately allocated costs is your finding to explain. The practices in our guide to monitoring subrecipients apply directly, and asking for a subrecipient's allocation methodology at the point of contracting is far easier than asking for it after their invoices have already been paid.
Conclusion
A cost allocation plan is a modest document doing serious work. It takes an organization whose costs blur across programs and funders and produces a defensible account of who paid for what and why. The regulatory requirement is not onerous: allocate according to relative benefit, use a reasonable base, document it, and apply it consistently. The difficulty is almost never conceptual. It is that writing the plan is never urgent, the person who should write it is the busiest person in the building, and nothing forces the issue until a funder or an auditor asks.
The organizations that handle this well share a few habits. They match each base to the pool it measures rather than applying one base everywhere. They capture actual effort rather than charging budgeted percentages for twelve months. They true up quarterly and leave a dated record of each true-up. They keep the plan, the audited financials, and the Form 990 flowing from one methodology instead of three. And they review the whole thing annually, because a plan that accurately described the organization four years ago is now a liability wearing the costume of a control.
AI shortens the parts of this that have kept plans unwritten: the drafting, the tedious comparison of allocations against activity data, the trend analysis that reveals drift nobody noticed, the reviewer questions you would rather hear now than in an audit room, and the small documents that build the trail. What it cannot do is choose your bases, interpret your awards, or stand behind the numbers. The finance lead owns the policy and owns the figures. Use the tool to get the plan written and pressure-tested, and keep the judgment where the accountability already lives.
Put Your Allocation Methodology in Writing
We help multi-program nonprofits build cost allocation plans that hold up under audit, with AI doing the drafting, the comparison, and the documentation trail.