Most researchers write the data management plan last, the night before submission, and they almost never put a dollar figure next to it. Then the award arrives, the plan promises a repository deposit and a cleaned, documented dataset, and there's no money or time set aside to do either.
Funders expect you to budget for this work, and in most cases they'll pay for it. This guide covers what counts as a data management and sharing cost, how NIH and NSF want it shown, a simple way to estimate it, and the mistakes that get costs cut.
What counts as a data management and sharing cost
NIH spells this out more clearly than anyone else, in NOT-OD-21-015. Allowable costs fall into a few groups:
- Curating data and writing documentation. Formatting data to community standards, de-identifying it, writing metadata and a codebook, and preparing files for a repository.
- Local data management. Specialized storage or infrastructure you need before the data go to a repository, when it's specific to the project and not something your institution already provides.
- Repository fees. Deposit or preservation fees, including fees for more than one repository if your plan names more than one.
What isn't allowed: infrastructure already paid for through your institution's indirect (F&A) rate, like the campus network or the general file server, and the ordinary costs of doing research. Collecting and analyzing data is research. Cleaning, documenting, and depositing it so someone else can use it is data management and sharing.
NSF's rules are less itemized, but its proposal guide treats the costs of documenting, preparing, and sharing research data as allowable direct costs, as long as they're explained in the budget justification. Other federal agencies and many foundations work the same way. Check the solicitation for anything specific.
How NIH wants the costs shown
Until 2023, NIH asked for a single "data management and sharing costs" line. That requirement was dropped for due dates on or after October 5, 2023 (NIH's announcement). Now you put each cost in its normal category and explain it in the justification:
- A research assistant's hours spent on curation go under personnel.
- A repository fee goes under other direct costs.
- Project-specific storage goes under equipment or other direct costs, depending on what it is.
Then the budget justification (or the additional narrative justification for modular budgets) gives a short breakdown of the estimated data management and sharing costs, so a reviewer can see the total. One or two sentences per item is plenty.
One NIH detail catches people: if you promise to keep data in a repository that charges a fee for, say, ten years, the whole fee has to be paid before the award ends. You can't budget for year eight of preservation in year eight.
A simple way to estimate
You don't need a cost model. Start from the promises in your data management plan and put a number on each one.
- List the outputs you'll share. The final analytic file, the raw data, code, instruments, interview transcripts. Each one has its own preparation work.
- Estimate hours per output. Writing a codebook, running a de-identification pass, cleaning and checking files, writing a README, and preparing the deposit. For a modest survey dataset this is often a few days of work, not a few hours. Qualitative data takes much longer, because transcripts have to be read line by line.
- Assign the hours to a person. Usually a graduate or research assistant, sometimes a data manager. Multiply by their rate plus fringe, and put it in personnel.
- Look up repository costs. Many generalist repositories are free up to a size limit, and some disciplinary archives charge for curation or restricted access. Check the repository's current pricing page, since fees change. The repository guide explains the options.
- Add storage only if it's really project-specific. Large imaging files, a secure enclave for sensitive data, or cloud storage the campus doesn't provide. If central IT already covers it, leave it out.
- Put the hours in the right year. Most sharing work happens in the final year, when the data are final. Budget it there, not spread evenly.
Here's the shape of a justification entry. The numbers are made up; use your own.
| Item | Category | Example justification |
|---|---|---|
| Data curation | Personnel | Graduate assistant, [80] hours in Year 3, to clean, de-identify, and document the survey data and write the codebook for deposit. |
| Repository deposit | Other direct costs | [Repository name] curation fee of $[amount], paid in Year 3, covering preservation for the retention period in the plan. |
| Secure storage | Other direct costs | Encrypted project storage for identifiable interview audio, $[amount] per year, not covered by institutional F&A. |
Mistakes that get costs cut or questioned
- A plan that promises more than the budget pays for. If the plan says "all data will be deposited with full documentation" and the budget has nothing for it, reviewers notice, and so does your program officer at closeout.
- Charging for things the institution already provides. General storage, backup, and software licenses the campus already pays for are usually part of F&A.
- Vague justifications. "Data management: $5,000" invites a cut. Hours, who does the work, and what it produces are what make it defensible.
- Forgetting the timing. Sharing costs in Year 1, when there's no final dataset, look odd. So do preservation fees that would need to be paid after the award ends.
- Leaving your sponsored programs office out. They know your institution's rates, which costs are already covered, and how your funder's forms work this cycle. Ask them early.
If you support researchers
A lot of this is the same from proposal to proposal: the list of allowable costs, the repository fees, a few sample justification sentences. A one-page budgeting guide from the library or research office, kept current, saves every PI from working it out alone, and it pairs well with a DMP checklist. If your campus is setting up that kind of help, the guide to starting research data services is a good first read, and I work with institutions on exactly this. Get in touch if that would help.
Common questions
Can I include data management costs in an NIH modular budget? Yes. Modular budgets still include these costs within the modules, and you describe the estimated data management and sharing costs in the additional narrative justification.
Are repository fees allowable on NIH grants? Yes, if they're needed to share the data as described in your plan. Fees for preservation over several years must be paid before the award period ends.
Can I charge a data manager's time to the grant? Yes, for the time they spend on this project's data management and sharing, charged under personnel and explained in the justification. General institutional data support is usually covered by F&A instead.
What if my data will be shared for free? You still have preparation costs. Cleaning, documenting, and de-identifying data takes staff time even when the repository costs nothing, so budget the hours.