Alongside millions of new research publications each year1 is the creation of millions more laboratory notebook entries. These contain important metadata, reflecting the nuance of experimental work. The ability to readily access, use and share laboratory notebook data allows researchers to quickly infer meaning from results and can help facilitate reproducibility across experiments. Collaborative or multidisciplinary research fields require efficient methods for capturing and sharing notebook entries between a diverse range of scientists.

Research relies on computing to analyze and present data; therefore, storing laboratory notebook entries in a digital format allows them to sit seamlessly alongside research data as they are processed. Electronic laboratory notebooks (ELNs) are fundamentally a means of digitizing entries at the point of creation, enabling those data to be processed computationally. However, they are not a panacea. Before deploying an ELN, it is critical that the requirements of users, as well as the advantages and disadvantages of different approaches, are properly understood to avoid creating a system that hinders rather than helps.

The past 20 years have seen a rapid increase in the number of ELN software packages

ELNs have been mooted in various forms since the late 1950s2. In the 1980s, software such as RS/1 (Bolt, Beranek and Newman, Inc.) offered researchers the capability to store, analyze and comment on data3,4. ELNs are presented as a tool for improving the reproducibility of research by facilitating the transfer of vital experimental details, both between generations of researchers and across different research groups5,6. Recording, accessing and preserving paper-based records can be slow, inefficient and difficult to integrate with modern computer-controlled data capture systems. However, implementing an ELN is non-trivial. Its adoption requires clear understanding of notebook use in a given laboratory setting and the provision of sufficient resources.

Most current ELNs are commercial offerings. These offer access to proprietary software, typically hosted remotely and available via subscription, under a software-as-a-service (SaaS) business model. A few community-developed open-source ELNs exist, with freely accessible codebases. There are also a small number of commercial ELNs with open-source codebases and free (to non-profit organizations) ELNs with proprietary codebases. Reviewing specific products is beyond the scope of this article; however, a number of product comparisons are available online7,8,9,10.

In the past 20 years, the number of ELNs has increased dramatically, as the benefits of digitization have been recognised (Fig. 1). In this marketplace, not all ELNs have proven successful. A significant number of both commercially available and open-source software packages have ultimately become defunct. In our analysis (Supplementary Method) of 172 ELN products (96 active and 76 defunct), the average lifetime of an ELN was found to be 7 ± 4.4 years (median ± consistent median absolute deviation). The lifetimes break down as 6 ± 4.4 years (n = 25) and 7 ± 4.4 years (n = 147) for open-source and proprietary codebases, respectively. The longest-running open-source ELN in our survey was ELOG (Stefan Ritt, Paul Scherrer Institut)11, which has been active for 20 years. The longest-running proprietary ELN that we found was Gene Inspector (Textco BioSoftware, Inc.)12, which has been active for 25 years. Company acquisitions, changes in the commercial market and lack of developer support or funding for open-source projects can all result in defunct ELNs. Long-term support and data access should be a primary concern when implementing an ELN: many benefits (e.g., rapid access to historic notebook entries) disappear if archived material is trapped inside inaccessible legacy systems, or worse, deleted. Procedures for extracting and archiving data in accessible formats should be part of any deployment strategy.

Fig. 1: Timeline of 172 ELN software packages, documenting known or estimated release and cessation year.
figure 1

The areas of the circles shown at the top of the figure are proportional to the numbers of new ELNs launched in a given year. Data are segregated into proprietary (blue) and open-source (pink) codebases and sorted within these categories according to the number of years that the software has been active. Insets are screenshots of a selection of ELNs, showing the progression of software development with time, including: RS/1 (Bolt Beranek and Newman Inc.), adapted with permission from ref. 3, ACS; ELOG (Stefan Ritt, Paul Scherrer Institute), reproduced with permission from ref. 11; eLabFTW (Deltablot), reproduced with permission from ref. 24; OSF (Centre for Open Science), reproduced with permission from ref. 58 under a CC-BY 4.0 license. See Supplementary Method for a description of the survey methodology and limitations. This plot incorporates data from primary and secondary sources9,14,59. An interactive version of this figure, along with any updates, is available from a data repository (Zenodo) supporting this paper52.

Before choosing an ELN, the purpose of the laboratory notebook must be identified

A laboratory notebook serves various purposes. For the researcher, it is a record of experiments and work conducted. The notebook may describe experimental methods, be a direct record of original data or provide metadata required to contextualize other data. Formal metadata (experimental test parameters and control conditions) may be supplemented by unplanned observations and annotations, both facilitating data analysis and interpretation. A notebook may journal both the genesis of ideas and the decision-making process13. Successfully capturing this information is critical to the researcher and others attempting to replicate the work.

Laboratory notebooks can be used to enforce good practice and to standardize workflows. For example, institutions may mandate the inclusion of risk-assessment templates within synthetic chemistry notebooks to encourage researchers to identify and mitigate hazards immediately before conducting a reaction. Routine procedures with well-defined outputs may follow a standard notebook template, to streamline information capture and standardize record keeping or to record quality-control procedures. This process can aid adherence to best research practices, such as ensuring that enough details are captured to facilitate reproducibility. An ELN can simplify this by acting as a database of templates and protocols14,15. For researchers, managers and institutions, laboratory notebooks provide evidence of work completed, facilitating internal accountability and providing a legal record to demonstrate regulatory compliance and potentially aiding intellectual property protection.

Identifying a given laboratory’s requirements defines and constrains the choice of ELN. Academic research laboratories typically feature a diverse range of experiments, data types and disciplines, resulting in users having different requirements from the same ELN package16. Although many ELNs are specialist products targeting researchers in a specific domain (e.g., biochemistry or pharmacology), these may not be relevant or sufficiently flexible for most researchers. Although this limitation was recognized by the early creators of ELNs in the 1990s17, it remains an issue that has been highlighted again in recent studies14. Record keeping may be required to meet regulatory standards, for example, for laboratories accredited to the testing and calibration standards ISO 17025:2017 and ISO 15189:201218,19, which stipulate requirements for laboratory information management, or for those wishing to adhere to general electronic record-keeping standards such as the Code of Federal Regulations Title 21 Part 1120. The motivating factors and requirements for a research group may not be the same as those of other groups at the same host institution, so care must be taken to identify the priorities of different stakeholders before selecting a particular product.

ELNs recover researcher time and enable better research practices, in return for financial cost

ELNs provide quality-of-life improvements over paper notebooks. In environments where ELNs have been implemented, the ease with which information can be sought and shared is regarded as one of the key benefits15,21. Figure 2 illustrates different ways that information can be shared via an ELN. Making ELN entries visible to multiple users is often simple to accomplish within the software. Project teams can instantly access relevant experimental data from different researchers, facilitating project management. Supervisors can remotely provide feedback without physical access to a notebook and add digital signatures to verify entries. Collaborators can be geographically separate, working in separate laboratories across multiple countries. Similarly, researchers can access their records from multiple locations, for example, from different laboratories or from home. This can be an advantage where physical access to facilities is restricted, as seen during the coronavirus disease 2019 (COVID-19) pandemic. It also mitigates the need to transport physical laboratory notebooks between locations, reducing the risk of cross-contamination and data loss.

Fig. 2: Examples of possible information flows between different users, as mediated by an ELN.
figure 2

Researchers can create and secure entries, share information with colleagues and collaborators and access the records of former team members. The degree of sharing is dictated by local policy, software features and configuration.

Notebook entries can be archived in situ as team members leave, while allowing incoming students and staff instant access. The ability to rapidly search through all available content allows researchers to filter and access ELN entries pertinent to their own work13,21. When users leave an organization, if permitted, an ELN allows them to create copies of their entries for future reference. This can allow quick access to previous and ongoing research ideas as a researcher progresses through their career. Physically storing and preserving digital data over long periods is more space and time efficient than attempting to store paper notebooks.

Table 1 illustrates how the differences between interacting with paper and electronic laboratory notebooks are ultimately a choice between time and money. Paper notebooks can be an inexpensive medium, but executing actions that are trivial with an ELN (e.g., searching, sharing and data backup) are time expensive. Conversely, most ELNs are relatively expensive to implement and maintain, compared to paper notebooks, but provide far greater functionality at a much lower time cost to the researcher. Actions such as searching, reordering, sharing and archiving can be extremely fast compared to paper notebooks. Depending on the implementation, ELNs may introduce a time delay for users because they require turning on hardware and authenticating into the software before a notebook entry can be made. The degree to which this is an issue depends on the hardware (a modern tablet can wake and unlock within 1 s) and software (biometric authentication can offer rapid logins, or, more commonly, local policy can dictate how long user sessions remain active before forcing reauthentication).

Table 1 Various laboratory notebook interactions, comparing their ease between paper and electronic media

The operating environment affects both paper and electronic notebooks. Laboratories that contain some form of protected environment (e.g., cleanrooms or biological containment laboratories) may have restrictions on the movement of items into and out of the space. Both paper notebooks and computer hardware can be contaminated in the laboratory. An ELN may help alleviate these issues by allowing access to notebooks via devices that remain inside the protected environment. However, this requires pre-planning of hardware requirements.

When considering any ELN system, the benefits of enabling more time to be dedicated to research and improving knowledge transfer and experimental reproducibility are balanced against financial costs. Recognizing who will bear these costs is important. For example, relying upon users to provide their own computing device to access an ELN effectively transfers this cost onto the researcher. This may cause people to spurn ELN adoption and disadvantages researchers without existing devices, as seen in studies of undergraduate web-based learning technologies22,23. If a laboratory already has a sufficient number of network-connected workstations, then this cost may be nil; otherwise, it can form a significant proportion of the overall implementation cost (a factor that is not included in any software vendor pricing).

An ELN is typically a combination of a user interface, a centralized database and a file store

Figure 3 illustrates a simplified view of a commonly adopted ELN architecture (although many different approaches are possible)11,24,25,26,27. Notebook entries may be stored in a relational database with attached files in a file store, with the ELN software facilitating user access and defining how notebook entries can be written and read. The software may be accessed via a web browser or in some cases through a custom application written for a specific platform (e.g., desktop or mobile operating system apps). Application-based ELNs may cause compatibility issues in academic research environments that typically feature a diverse range of operating systems28. Depending on the implementation, the separate ELN components may be separate servers27, in different physical locations. This major conceptual shift from paper notebooks brings both opportunities and challenges. Although the underlying technology is ideally invisible to the end user, the choice of ELN can influence the availability of different features. For example, most ELNs are ill suited to storing large volumes of raw data (either from a performance or cost perspective). A locally hosted ELN server may rapidly run out of physical storage space if not appropriately provisioned. A cloud-hosted server may quickly incur significant hosting costs as storage and bandwidth requirements increase. Even with these restrictions, the amount of information that can be stored is more than is possible with paper notebooks, which can store only small quantities of data.

Fig. 3: Simplified illustration of a commonly used ELN architecture.
figure 3

Users interact with the ELN software via workstations and/or mobile devices, with off-premises access possible via technologies such as virtual private networks (VPNs) or by exposing the ELN to the wider internet (the latter introducing additional security concerns). Experimental hardware may interact directly with the ELN server (e.g., via an application programming interface (API)). The ELN software, database and file storage may coexist on the same physical server or operate on separate hardware. The server may be run as a local installation at an institution or as part of a cloud-hosted product from the software vendor. Additional requirements include backup servers, which should be geographically separate from the ELN (and should consist of multiple layers of redundancy). Third-party services include access to single sign-on servers (to facilitate a better user experience), trusted third-party timestamping authorities (for independently digitally signing notebook entries) or other laboratory systems (such as laboratory inventory management systems).

Sensitive data (e.g., patient or commercially sensitive information) may fall under local institution or legally mandated data-protection regulations (e.g., GDPR 2016/67929, which dictates the handling and safeguarding of personal data within the European Union). This may restrict the physical location and transfer of data, excluding the use of off-premises ELNs that use international cloud-based infrastructure30. Locally hosted ELNs may provide greater control over data security by restricting notebook access to users inside an institutional network (i.e., access devices must be physically on the premises or connected to the internal network remotely, for example, by using a virtual private network).

Data integrity can be enhanced by using version control and timestamping

The ability to create immutable notebook entries that cannot be removed or altered after creation is critical for academic and legal integrity. Paper notebooks typically implement this through rules and procedures (e.g., by prohibiting the removal of pages and signing and counter-signing entries). Not all ELNs address this issue, again precluding the use of some products. For example, general note-taking software packages, such as Microsoft OneNote and Evernote, do not typically include features to digitally sign or timestamp entries, with workarounds such as signing exported files required28,31. The level of verification required depends on both regulatory requirements and locally accepted laboratory notebook standards. For example, it may be sufficient to rely upon administrator-defined software features, such as the ability of a supervisor to lock notebook entries to prevent them from being edited or to disable entry deletion. Many ELN packages include a mechanism for version tracking, which record how a document is edited over time, a potential deterrent to modifying entries after the fact. This is a conceptual difference between paper and electronic notebooks. In a paper notebook, entries are instantly recorded (presuming some form of indelible ink is used). In an electronic notebook, there is typically a finite amount of time during which the entered text or other content remains malleable before it is saved to the server. This period can range from seconds to minutes, with the server creating intermediate versions of the entry, or a longer period until the entry is locked or finalized through some technical means. Excessive versioning may significantly increase storage requirements and overall running costs; hence, local policy is required to determine a suitable compromise for the period of time between making an entry and some form of versioning and ultimately finalization.

If verifying the provenance of notebook entries is critical, the ELN should incorporate technologies that adhere to recognized standards for trusted timestamping, such as RFC 316132 or ANSI X9.95-201633. Trusted timestamping uses an audited third-party organization (typically a commercial provider) to digitally sign and timestamp a file. A cryptographic hashing algorithm is applied to a digital file (e.g., a portable document format file that corresponds to a notebook entry. The algorithm generates a file containing a hash (a mathematically unique representation of the original data). This hash cannot easily be reverse-engineered to recreate the original data and thus can be safely transmitted to the trusted timestamping authority via the internet. The authority digitally signs and timestamps the hash, in the process generating a third file (a timestamp token). This token is sent back to the ELN software and stored alongside the original portable document format file. Any modifications to the original file will invalidate the token, because recalculating the hash of the modified file will result in one different from the value contained within the digitally signed token. The process allows a digital file’s contents at a given point in time to be verified. Although technically complex, this process is typically performed invisibly to the user.

While trusted timestamping incorporates the concept of digital signatures, it adds the additional benefit of not only verifying the signer’s identity but also the time at which an entry was timestamped. In practical terms, local policy is required to ensure that users routinely timestamp their notebook entries, as the date and time of timestamping are being recorded, not the instant when an experiment was conducted. Procedures for archiving data should ensure that both the timestamped file and the timestamp token are properly preserved.

Open-source and commercial ELNs have different costs and benefits

In our survey, commercial ELNs were observed to be far more prevalent than open-source ELNs (147 versus 24 products identified). Table 2 compares commercial to open-source approaches. Although SaaS is a widely adopted licensing model (e.g., institutional subscriptions to Microsoft Office 365 or Google Workspace), SaaS ELNs may be prohibitively expensive for individual research groups because of per-user pricing, ever-growing file-storage costs and an indefinite subscription required to maintain access. In 2017, Kanza et al. reported that, in a survey of 169 users participating in an ELN pilot study, both limited financial budgets and the time required to implement an ELN were major concerns14. Similarly, although open-source ELNs can have relatively low initial and ongoing financial costs, they require time to run and maintain and may require server hardware to be purchased. The relatively modest requirements for many open-source ELNs make it feasible to repurpose old computing hardware to act as an ELN server, helping to reduce this cost. Costs can scale with the size of deployment; for example, commercial site-wide licenses may be negotiated at preferential rates rather than the per-user pricing available to individual research groups34. Implementing an open-source ELN at institutional scale can take advantage of pre-existing server infrastructure and support from information communication technologies departments. Some commercial providers offer free or reduced pricing for academic users14.

Table 2 Comparison of ELN features for open-source and commercial solutions (note that this is a generalized comparison—not all ELN products offer the same levels of functionality)

Open-source ELNs have the benefit of allowing not only the data to be archived, but also the underlying software itself. Technologies such as virtualization and containerization present feasible pathways to preserve the operating environment of the ELN for future access, beyond software and hardware obsolescence. For example, provided the codebase has been properly archived, open-source ELN software may be resurrected as a virtual machine within a modern operating system, to allow historic file access and export. This can help facilitate future file interoperability by providing easy access to the original software. The open codebase also means that data formats and standards are fully exposed, facilitating the future development of tools to reparse or extract data. When assessing an ELN hosted by a third party, consider what will happen to notebook entries when the product is discontinued. Some open-source projects (such as the Open Science Framework)35 may include contingency plans and funds to ensure the ongoing preservation of research data. If an ELN provides data export functions, these should be tested to confirm that they provide the required level of functionality. For commercial ELNs, it should be ensured that supplier contracts include the necessary terms to facilitate data export.

ELNs are not a solution to poor data management

ELNs do not resolve the challenge of systematically storing raw data and are just one part of a holistic data management strategy. Successfully implementing an ELN requires reflection on current practice to determine how a laboratory handles and stores information36,37. As with paper notebooks, policy, training and enforcement are required to ensure that users record timely, useful and complete notebook entries. New users require training to understand the purpose and expectations of laboratory notebook use within a given organization. A policy for how, when and by whom notebooks are monitored sets expectations. An offboarding procedure should be implemented to ensure that outgoing user data are appropriately archived.

Procedures for linking raw data, laboratory notebook entries, analyzed data and publication data should be clearly defined and enforced. Data should be stored on centralized group or measurement-specific servers or publicly available repositories, with redundant backups. Persistent identifiers such as digital object identifiers can be used to help link resources, with ELN entries acting as centralized documents that connect to multiple files and data36. Some ELNs already generate unique identifiers for each notebook entry24. To help verify the integrity of externally stored data, ELNs can be used to record cryptographic hashes of data files.

Some ELNs feature application programming interfaces that allow other software to directly read and write notebook entries24. For example, a user conducting a computer-controlled experiment could allow that equipment to directly record experimental or process details, potentially streamlining routine measurements. Laboratories with well-defined workflows, such as electron microscopy38 or genetic analysis27, may benefit from specialized ELNs that incorporate notebook entries within the data capture workflow or that have been designed with equipment integration as a primary goal.

Successful ELNs recognize the needs of their users

Within an academic research environment, both time and money are at a premium. Ideally, if adopted, ELNs for the academic research laboratory should be implemented at the institutional level, harnessing existing ICT infrastructure, and with a sufficient commitment for ongoing support to encourage uptake39. Although a handful of university-based surveys and studies of ELN implementation exist14,39,40, the reported level of success varies, suggesting that careful user engagement, product choice and ongoing support are key to successful deployment.

Critically, the implementation of an ELN should not introduce a net burden to the user. If users are unable or unwilling to use time-saving features, adopting an ELN may ultimately be a hindrance. For example, for a researcher who regularly draws chemical structures or writes equations into their notebook, most non-specialist ELNs will be less convenient than paper39, and the introduction of an ELN may be undesirable16. Lack of support for systems such as LaTeX can be a barrier to adoption in specific disciplines41. Recognizing the needs of an often-diverse range of researchers is essential before making decisions on the choice of software and approach.

Internal trialing of a small number of products, jointly agreed compromises or incorporating integrations with specialist solutions, such as ELNs capable of handling chemical structures42, may be required. Simple infrastructure factors, such as unreliable WiFi7,14,23, insufficient benchspace to place a laptop computer14 or lack of access to up-to-date hardware and software in the laboratory, can also adversely affect users39. The prevalent culture of private and personal academic laboratory notebooks should be recognized, with one ELN study noting that researchers felt embarrassed when required to share their notebook with colleagues43.

User training should be recognized as an additional time burden, with researchers already expected to master a wide range of software packages44. Similarly, many laboratories already use laboratory information management systems for inventory handling, equipment booking and procurement. Introducing a further standalone system can work against the time benefits of digitization and integration45. Implementing an ELN is an opportunity to reflect upon and consolidate existing practice, with many ELNs incorporating laboratory information management system–style components. Similarly, if it is intended for the ELN to integrate with specific pieces of hardware, it cannot be assumed that software will work seamlessly; thus, system integration should be tested before deployment, and the appropriate resources should be allocated to maintain the integration.

To date, there have been no published multi-year longitudinal studies of ELN implementation in academic environments. Hence, it is important to consider user issues that may appear over longer time scales. For example, determining how future researchers will be made aware of existing records if the original author (or the author’s supervisor) has left the institution. Individuals should be identified who will facilitate access requests to existing records from new users. Investing in user training from the start should ensure that third parties are able to effectively locate these records in the future (i.e., that sufficient notation and metadata are being captured at the point of entry). Academic institutions may be able to take advantage of the existing in-house expertise of academic librarians, research data managers, compliance administrators, systems administrators and archivists to help develop long-term workable policies. The general challenge of long-term digital data preservation and access exists within any large organization (e.g., email preservation), so leveraging existing good practice may aid deployment.

Box 1describes our own experience of implementing an open-source ELN, illustrating some of the points discussed in the article. The choices made arose out of the specific set of circumstances appropriate to our research group; thus, this should be considered not as a prescription for the best approach to implementing an ELN but rather as a reflection on what we have learned.

ELNs are an opportunity to consider the broader philosophy of academic laboratory research

With care, ELNs can be used to support information capture, making it more consistent, accessible and usable to both current and future generations of scientists. Implementing an ELN provides an opportunity to consider how digitized notebook entries can help address some of the broader challenges of academic research. For example, some ELNs cater to aspects of open science46,47, allowing them to be configured to share data outside of organizations, supporting initiatives such as the FAIR Data Principles (a proposal that scientific data should be findable, accessible, interoperable and reusable)48. Similarly, the integration of ELNs with institutional repositories may offer new opportunities to tackle the challenge of research data management, such as efforts by the University of Edinburgh to allow users to directly deposit data into an institutional repository via their ELN interface49.

Adopting an ELN abstracts the user from the underlying notebook storage technology, allowing a wide range of other approaches to be implemented, different from the one illustrated in Fig. 3. Rather than using a private database of notebook entries, an ELN could implement a publicly distributed, decentralized record store, by using technologies such as blockchain and peer-to-peer networking50, to aid accountability or reduce the reliance on only one repository for long-term data storage. For example, the research project bloxberg is a blockchain operated by an academic consortium51 that has been independently integrated into the ELN eLabFTW24. Cryptographic hashes of notebook entries can be timestamped and recorded on a publicly accessible ledger, distributed across the consortium network, thus removing the need for a single trusted third-party timestamping authority. In effect, verification of the integrity of notebook entries (or other scientific data) is distributed among the consortium members.

Because ELNs provide the primary interface to research data, there is the opportunity to consider how they can integrate with computational approaches that aim to automatically derive insights into the data. This could include the integration of an ELN with computational semantic technologies16, which allow the meaning of human language to be automatically inferred. This would allow research metadata to be automatically derived, aiding search efforts, or create automated insights by linking relevant data together45. Alternatively, an ELN might be integrated with existing community-defined ontologies and databases for specific disciplines. In this scenario, the ELN would act both as a form of input validation, ensuring that data are captured according to community standards, and as a mechanism for facilitating access to underlying data to facilitate large scale meta-analyses.

In summary, for researchers and institutions considering implementing an ELN, a nuanced understanding of laboratory culture is needed to facilitate a respectful and ultimately user-supported deployment. The barriers to entry must be carefully managed because these have the potential to create technological divides within the academic community, diluting the benefits of ELNs. Successful ELN implementation should be seen as an ongoing commitment to ensure that the needs of different users continue to be met. Given that the current median lifetime of ELN software packages is 7 years, it is of utmost importance to have sufficient ongoing institutional support to maximize the value to researchers and mitigate the risks, enabling continuity when software changes are required. With careful consideration, successful implementation of an ELN presents a pathway to greater knowledge development and transfer within academic research.