Best Practices for Storing and Saving Research Data

Every research project, no matter how well designed, is only as good as the data it rests on – and only as reliable as how well that data is managed. Collecting information is just half the work. The other half is making sure your data is stored safely, organized logically, and attributed correctly so that when it comes time to write your report or share your findings, everything you need is exactly where you expect it to be. Poor data management leads to lost files, confused timelines, and even credibility issues. This post breaks down what researchers need to do to store and save their data properly – from the moment they begin collecting it to the moment they publish.

Why data storage is a research priority, not an afterthought 🔗

Many researchers treat data storage as something to worry about later. That’s a mistake. According to the University of Chicago Library, data is easily lost, digital files are fragile, and storage formats can become obsolete over time. A storage strategy needs to be in place before data collection begins – not after. There are two broad types of data in most research projects: primary data, which you collect yourself through surveys, interviews, observations, or experiments; and secondary data, which you source from existing studies, databases, government records, or published reports. Both types need careful, intentional management throughout the research lifecycle.

Choosing the right storage medium 🔗

Not all storage options are equally suitable for all types of data, and the choice has real consequences for security and long-term access.

Local and institutional storage 🔗

The most common starting point is your personal computer or laptop. It is convenient, but it carries significant risk. Washington State University’s Research Data Management guide points out that computer crashes, theft, and accidental deletion are real hazards, which is why you should never rely on a single copy stored on one device. Institutional or network drives – servers managed by your university or organization – are far more stable options. They are maintained by IT staff, have built-in security protocols, and typically include automatic backups.

External drives and cloud storage 🔗

External hard drives offer a practical backup option and many support encryption for sensitive data. However, research data management guidance from Washington State University advises against keeping external drives physically near your main computer – theft or disaster could wipe both at once. Flash drives and optical media like CDs or DVDs degrade over time and are not suited for long-term storage. Cloud services such as Google Drive or Dropbox offer convenience and remote access, but come with caveats: they are owned by private companies, may not guarantee long-term data security, and are generally not recommended for sensitive or confidential research data.

The 3-copy rule 🔗

Tennessee State University’s data storage guide recommends keeping at least two copies of original data – one stored locally and one stored remotely in a different geographic location. This redundancy ensures that if something goes wrong in one place, your data survives. The University of Cambridge’s research data team reinforces this, noting that your storage choice must match the security risk level of your data and that a backup plan is non-negotiable.

Organizing your data: folders, files, and naming conventions 🔗

Storing data is not just about where you save it – it is also about how you structure and label it. Poor organization can make data practically inaccessible, especially as a project grows or when you return to it months later.

Building a logical folder structure 🔗

The University of New Hampshire’s data management guide recommends against dumping all files into a single folder. Instead, researchers should create a hierarchy of folders and subfolders organized by type, stage, or project component – for example, separate folders for raw data, interview transcripts, survey responses, analysis files, and final outputs. The top-level folder should clearly identify the project name and date. Folder names should describe content, not researchers’ names, and the structure should remain consistent throughout the project.

Utrecht University’s data management guide adds an important safeguard: keep your raw data in a separate, protected, read-only folder and always work on a copy. This protects the integrity of the original data in case something goes wrong during analysis.

File naming conventions that actually work 🔗

A good file name should tell you what is in the file without opening it. The University of Michigan Library advises including a few pieces of descriptive metadata in every file name – such as date, project identifier, and data type – formatted consistently. Dates should always follow the YYYYMMDD format so that files sort chronologically. Names should be under 25-32 characters, use underscores or dashes instead of spaces (which many operating systems do not handle well), and avoid special characters like /, \, *, ?, and &. Version numbers should be added at the end (e.g., v01, v02) so you always know which file is current.

The European University Institute’s research data guide recommends thinking of your file naming strategy as a system – one that is documented so that collaborators and future users can understand it without having to guess. A simple README.txt file placed in the top-level project folder, listing what each folder contains and explaining any abbreviations or codes used, goes a long way.

Documenting and tracking your data throughout the research process 🔗

Organization is not a one-time activity. As your research progresses, your files change, and keeping track of those changes matters. The University of Washington Libraries recommends implementing version control – tools like Git allow you to track changes in data files and code, which is especially valuable in collaborative projects. For every significant change to a working file, save a new version rather than overwriting the previous one.

Beyond file versions, documentation should capture the broader context of your data. The University of Guelph’s data organization guide outlines what effective documentation should include: descriptive information about the dataset, the research purpose, data creators, naming conventions, variable definitions, and details on how data was collected and processed. This information is what makes your data understandable not just to you, but to anyone who may access or review it later.

Storing and attributing secondary data correctly 🔗

When your research draws on secondary sources – published datasets, government statistics, journal articles, or archived records – managing that material requires an extra layer of attention: proper attribution.

Tracking your sources from the start 🔗

The moment you use data from an external source, record it. Using a reference management tool such as Zotero, Mendeley, or EndNote helps you maintain a running library of all sources consulted – including books, journal articles, datasets, and online databases. This prevents the scramble of trying to reconstruct your source list when writing up your report. Library cards, borrowing records, and downloaded PDFs should all be filed in a dedicated secondary sources folder within your project directory.

Why data citation is non-negotiable 🔗

Cornell University’s Data Services team defines data citation as the practice of referencing data products used in research, including key descriptive information such as the title, source, and responsible parties. Properly citing your data sources creates a bibliographic trail that connects your publication to the data underlying it – establishing a clear and verifiable timeline of how and where information was used.

Temple University’s research data management guide explains that data citation gives credit and attribution to the original creator, encourages sharing and re-use, enables verification of research results, and helps track the impact of data over time. A complete data citation typically includes the author(s), year of publication, title of the dataset, publisher or repository, and a persistent identifier such as a DOI.

The University of York’s research data guide notes that the same principles of attribution applied to books and journal articles apply equally to research datasets. When you reuse data shared by another researcher or organization, you are legally and ethically required to provide proper attribution – failure to do so can constitute plagiarism or a copyright violation.

Using DOIs and persistent identifiers 🔗

A Digital Object Identifier (DOI) is a persistent, unique identifier assigned to a digital object – a journal article, a dataset, a report. Including DOIs in your citations means that even if a source’s URL changes, the link remains valid. Research published in the Data Science Journal highlights that while researchers are beginning to use DOIs to reference data, formal data citation is still not consistently practiced – and vague or incomplete references make it harder for future researchers to verify findings. Getting into the habit of recording DOIs from the start saves significant time and protects research integrity.

Security and access control 🔗

Stored data must also be protected. The University of Washington Libraries’ data storage guide outlines a layered approach to security: use strong passwords and multi-factor authentication for accessing data systems, implement role-based access controls so only authorized team members can view sensitive files, and anonymize participant data wherever possible to minimize risks in case of a breach. Password managers are recommended for keeping credentials secure without losing them. Devices that contain research data should also be encrypted, particularly laptops and portable drives that could be lost or stolen.

Planning data management from day one 🔗

The most effective data management happens when it is planned before a project begins, not improvised along the way. A Data Management Plan (DMP) is a formal document that outlines how data will be collected, stored, organized, backed up, secured, and eventually shared or archived. A peer-reviewed guide to effective research data management published in PMC notes that policies, guidelines, and compliance measures shape how researchers work with data at every stage – from collection through storage to publication. Many research funding bodies and academic institutions now require a DMP as part of any research proposal. Tools such as DMPTool can help researchers create structured plans that satisfy these requirements.

What do you think? If a researcher loses months of interview recordings because they only saved one copy on a laptop that crashed, whose responsibility is that – the individual researcher, the institution, or both? And how early in a research project do you think data management planning should begin – before data collection, during, or does it really matter as long as it happens eventually?

References

1. https://guides.lib.uchicago.edu/datamanagement/storing
2. https://libguides.libraries.wsu.edu/rdmlibguide/datastorage
3. https://tnstate.libguides.com/c.php?g=1149650&p=8949924
4. https://www.data.cam.ac.uk/data-management-guide/looking-after-your-data
5. https://libraryguides.unh.edu/datamanagement/organize
6. https://www.uu.nl/en/research/research-data-management/guides/storing-and-preserving-data
7. https://guides.lib.umich.edu/datamanagement/organize
8. https://eui.libguides.com/research-data-guide/document-organise-store/file-naming-folder-structure
9. https://guides.lib.uw.edu/research/dmg/storage-security
10. https://guides.lib.uoguelph.ca/OrganizingYourResearchData
11. https://data.research.cornell.edu/data-management/storing-and-managing/data-citation/
12. https://guides.temple.edu/datamgmt/citingdata
13. https://subjectguides.york.ac.uk/rdm/citing
14. https://datascience.codata.org/articles/10.5334/dsj-2023-010
15. https://pmc.ncbi.nlm.nih.gov/articles/PMC12685206/
16. https://dmptool.org