Datashare redesign makes research tool more powerful, more accessible for all

For the past decade, ICIJ’s Datashare has been the gateway for hundreds of reporters exploring the more than 100 million leaked files at the heart of investigations like the Panama Papers, Pandora Papers, FinCEN Files and more.

Now, after 10 years of continuous improvements, the research platform has received a major upgrade, with a suite of new and updated features aimed at making the tool easier, faster and more accessible for a wider variety of users.

The two-year redesign project drew on feedback from dozens of journalists, culminating in a new interface that enhances functionality and a design system that allows future platform upgrades to be integrated easily and seamlessly. See what’s new.

Maintaining a mountain of data

Datashare allowed ICIJ to centralize all the highly sensitive documents from the biggest leaks in history in a secure place. Investigations like the FinCEN Files and Pandora Papers relied on Datashare to help journalists unearth secrets buried in complex sets of documents using powerful search functionality and AI-fueled features.

But with each new investigation over the past 10 years, our developers have made iterative updates to the free, open-source platform, often responding to requests from our users or to the unique needs of different types of data at the heart of our various investigations.

Over time, those improvements made the user experience complex. For journalists who spend hours buried in documents inside Datashare, they need simplicity and efficiency. Even the most tech-savvy reporters were missing as much as 20% of Datashare’s powerful features that weren’t visible enough in the user interface.

At the same time, we have heard interest from other user groups who could benefit from a tool like Datashare — users such as researchers, civil servants and private firms.

After 10 years, it was time for a comprehensive makeover.

Datashare for all

To redesign Datashare, we began with our users: what did they like, what did they need, and how could we make the tool more accessible and useful for more people?

The design phase included online user interviews with reporters from all around the world, focus groups with journalists to compare design options, consultation with an accessibility specialist to make sure the new interface complies with accessibility guidelines and online high-fidelity prototype testing of the final versions.

ICIJ’s training manager Jelena Cosic, center, at a training session with ICIJ partners.

During this process, we learned that even if journalists really love Datashare, when pushed they would become like friends picking faults with your ex after a breakup: some said Datashare was too slow, others were not aware of some key features, particular bugs were really annoying, document-opening animations were inefficient (journalists go through hundreds of documents per day). One user even confessed “you know, Datashare is a daunting place.” We knew we had work to do.

Good design allows users to easily and intuitively navigate software — either because functions are clear or, as Jakob’s law suggests, because buttons are where users expect them to be, based on their experience with other commonly used websites.

Part of the challenge in this process came from one of Datashare’s most important security features — we don’t collect any information about our users. We don’t know what devices they’re using, which buttons they mostly click on or any other contextual information. Without this data, we had to ensure we were employing best design practices to make Datashare work for all users regardless of who they were or how they were using to the platform.

We brought in an accessibility consultant to help: we increased buttons’ sizes, font sizes and contrasts, on top of working on a better keyboard navigation and labeling components so that screen readers better work on Datashare.

Each page was designed with information hierarchy and prioritization in mind: as components are bigger for accessibility reasons, there is less room on each page. As a designer, I worked, with users’ input, on deciding which features are more important or which could be grouped to reduce cognitive load. This was a big challenge to tackle on an application with multiple options and loads of information to display.

The result is a streamlined interface that is accessible and intuitive for more users, whether journalists, researchers or anyone else who might benefit from the tool.

Another challenge was responsiveness: most journalists spend hours on computers to explore thousands of pages from thousands of documents. But sometimes they access Datashare from smaller screens, or even their smartphones. This made it all the more important to manage the hierarchy of our components and the information we were displaying.

For consistency, we created an entire design system to house all the components of the Datashare user interface: buttons, search bar, cards, menu, tables, and more. This system keeps us consistent across pages and makes the platform more comfortable for users, who always see the same components in the application. It also helps future-proof the design for new features to be added in the coming months and years.

So… what’s new?

Let’s dive into Datashare’s new interface:

Navigation: the menu has more entries to directly reach all pages, saving a few clicks each time.

Document’s full screen view: you can expand the view of a document for a more comfortable reading experience and use a carousel with previews to navigate between documents

Datashare’s redesigned document view.

 

Settings: on all pages, you can customize what data is displayed (metadata for documents’ cards, data columns for tables, etc). Each investigation is different and sometimes, you want to see documents’ creation date while other times, the language or type of the document is more important

User data panels: the number of tags and recommendations are now grouped and summed up on top of a document and you can open respective panels to see more details on the left of the document

Filters: because the list of available filters has grown over time, it was time to group them into 3 collapsable categories: documents info, user data and entities

A more intuitive list of filters.

 

Batch search results: a new page now allows you to directly see the number of results per query, while before, you needed to use the filters one by one

Dark mode: you can now switch between light and dark modes

 

You can check out this new version on Datashare’s demo website where you can explore the documents from the LuxLeaks investigation.

What’s next: speech-to-text, structured content and much more

The new version of Datashare is now available. Even though the prototype was tested on dozens of beta users, we expect feedback from users who are going to work on their own documents with this version. As we’ve experienced for the past 10 years, each investigation is unique and requires special adjustments: whether it’s translations, file types, entities or the way the documents were originally scanned, each set of documents brings its own challenge. Datashare aims to help users get the best from their files in as many use cases as possible.

As an open source software, we share our backlog on Github and we welcome feedback and suggestions. Upcoming features in the pipeline include:

  • Advanced search form: we want to provide a quick, user-friendly form, instead of requiring users to utilize operators like AND, OR, NOT or others in the search bar
  • Split view: for PDFs, we’d like users to see both the extracted text and the original document on the same screen so they can quickly access the original page where their search term has been found
  • Shared saved searches: users should be able to share the searches that they found interesting with colleagues, as they do for batch searches
  • Speech-to-text: users will be able to ask Datashare to extract text from audio and video files
  • Structured content: we want to extract content from the documents in a better way by preserving the data’s original layout and structure. We created a user survey on structured content – please fill it in if you have time
  • Re-extract text from documents: if text recognition is imperfect on scanned documents, users should be able to flag those files for re-processing
  • Alerts: users will be able to set up alerts and receive notifications when new documents added to Datashare match specified search terms
  • Highlights: we have plans to let users decide the length and the number of highlights they see when they search for a name and quickly see it in context in the documents

Anyone can download Datashare for free and work with their own documents on their computer with the local version of the software. A server version can also be set up by advanced technologists to allow teams to collaboratively work on the same documents online. The older versions of Datashare are still available on datashare.icij.org. If you have any feedback or suggestions on this new version, please share them with us by writing at datashare@icij.org.

Continue Reading