In June 2020, at the height of the pandemic, the National Library of Australia (NLA) successfully released a redesigned, rebranded, rearchitected, transformed Trove.

Trove is a free resource to explore all things Australian. A single digital point of entry to more than 6 billion digital items, Trove connects you to digital collections from hundreds of Trove Partners including Australian libraries, museums, galleries, media, government, community organisations and more.

Program governance

Beginning as a bespoke NLA IT development in 2008 through to 2016, Trove’s infrastructure was funded from within the Library’s existing appropriation, with approximately one-quarter of digitised newspaper content funded by Partners.

The 2016-17 Budget provided the NLA with $16.4 million over four years, to digitise historically significant material and upgrade critical infrastructure, including $8 million allocated to modernise Trove. The aims of the Trove Program were to refresh the brand, improve user experience and enhance IT systems in order to attract a wider audience, and increase user discovery of collections at a lower cost per engagement for the NLA.

To deliver this transformational Program, we adopted a Prince2/ Agile approach in concert with the Digital Transformation Agency’s Digital Service Standard. The NLA’s Digital team were experienced with developing using Agile techniques – the Business areas less so.

To bridge this gap, we created non-ongoing Product Owner roles for each of the Program development streams: branding; Trove redesign; and Trove search and backend. Support for staff in new roles included bespoke training, a weekly product owner forum, hands-on mentoring in Jira, user story development, and involvement in daily cross-functional standups and story points estimation with developers.

User experience design

It was critical to the Program’s success that the redesigned brand and technology attracted new audiences without alienating Trove’s existing passionate user base and Partners. This included an army of ‘Voluntroves’. These volunteers are a powerhouse supporting the NLA’s digitisation programs and their open research.

For example, when old newspapers and gazettes are scanned and made digitally available on Trove the character recognition technology often produces unreadable text. ‘Voluntroves’ can ‘text correct’ the digital copies, basically providing a crowd-sourced transcription service, and making the content searchable and useable.

Since 2009 all the available crowdsourced activities equate to 50 years of volunteer’s time! As you can imagine, many of these users had spent years developing their knowledge and expertise in the Trove systems, so were leery of any changes to be made. The UX design and development approach balanced the needs of existing and new users.

At Program start-up an ‘atlas’ of the existing Trove site was commissioned to document the complex pathways that had developed over nine years of bespoke development. Audience analysis was undertaken to identify existing personas, and those of desired new users to fully reflect Australia’s diversity. More transparent user journeys were developed based on thousands of user surveys about Trove.

Once we understood our audience and journey mapping, we commissioned updated branding. The brand needed to reflect the collaborative nature of Trove as a free resource for all Australians to engage with their heritage. Development of the brand involved extensive user testing of the logo and design both within and without the NLA, with existing and new users, including culturally and linguistically diverse, Aboriginal and Torres Strait Islander groups and Partner organisations.

We developed a web-based prototype, first static, and then functional, in concert with the branding team. This was tested with 65 focus groups in cities and rural centres across Australia, to provide feedback on useability and the appropriateness of branding. We undertook in-depth useability interviews.

Technical solution

So how did the technology delivery support the brand and user experience? Trove is now dynamic, designed to rely less on inhouse IT staff and more on business area staff to provide access to Trove’s evolving collection material. This was achieved by introduction of a new Drupal Content Management System (CMS). Drupal gives NLA a great combination of stability and flexibility. It provides flexibility to build, with reduced licensing costs and the ability to host and support in-house.

As a result, we were able to customise the CMS to suit Trove’s functional and nonfunctional requirements, allowing tight integration with other Trove systems. The Trove application was redeveloped using the vue. js open-source front end JavaScript framework to provide a consistent user and brand experience.

Trove has also been tweaked to improve search results, and to automate some behind the scenes processes. Upgrades to infrastructure and software have increased the reliability and security of the service. Changes to search results included the addition of significant new web archive content, enriching Australian’s access to the history of the last two decades.

Millions of additional pages of Australia’s documentary heritage, including parliamentary papers, early colonial records, maps and materials relevant to Australian Indigenous peoples has been digitised and is freely available going forward.

Trove Partners have been delivered a new business intelligence service, which provides them with metrics about their contributions and collections to Trove. The IT solution has also improved information discovery, by refining relevance ranking in search results, optimising on site searching, indexing full-text for born-digital content, and optimisation of external search through improved Google seeding.

Pandemic

In February 2020 as we commenced the public Preview, the team couldn’t have imagined that a month later we would be in lockdown due to a pandemic. In March – three months before the June launch of a 3-year development – the Program had to quickly adjust to this new way of working from home.

The planned launch dates remained, with a soft launch on 22 June 2020 where the old and new sites were run in parallel, and a hard launch cutover to the new site on 26 June. However, the Communications and Marketing team had the mammoth challenge of replanning the launch.

Originally planned for a face-toface event at Parliament house with the Minister and media on hand, the team redesigned the launch strategy at very short notice, and whilst working from home. The Launch became a digital event hosted by Annabel Crabbe, with the participation of Minister Fletcher the NLA’s Director’s General and key Partners.

It was live streamed to Trove VIPs with the broader community accessing the event in their own time via Facebook. Social media was replanned in light of the pandemic and launch strategy. It was one of the more unusual COVID experiences for the team to watch as silent audience members to the streamed launch – well socially distanced of course.

Lessons

Whilst the Program successfully delivered within the funding envelope and exceeded its forecast minimum viable product, there were lessons learned arising from the tight integration of brand, Trove application and back-end IT development.

The Program set an early constraint that there was to be no rearchitecting of back-end Trove. However, the decision to rebuild the Trove application using the vue.js platform coupled with the decision to service the new frontend via a new API both resulted in the need to increase both the IT build time and resource estimates.

Similarly delays in the front-end design – arising from the under-estimated complexity of managing UX testing (including finding participants), associated re-work of the interactive prototype, and knock on delays to digital Brand designs – created more IT integration work.

Integration required highlighting the importance of a formal Agile “IT Runway”, which was subsequently created to formally identify and schedule all the technical IT dependencies.

Outcomes

After a peak in July 2020, end user queries reduced below prelaunch levels. Key indicators show that the new user experience has indeed attracted new audiences and held them for longer. Average time on site increased by 30 seconds to 7:40, pages viewed increased by 0.5 pages to 8.9 pages per visit. Importantly the ‘bounce’ rate has dropped significantly from 52% to 30% (these are people that come & leave straight away); this may be a reflection of both Google filtering out automated traffic as well as user improvements.

Trove is seeing an increase in return users. In the 3 months pre redesign to 3 months post redesign we saw 1.5 visits per user and this has increased to 1.75 visits per user. The breakdown of users has also changed – 16% returning users to 18% returning users.

These numbers in combination start to show that Trove is engaging with our users more and first time users becoming regular users. Even the 2% increase in returning users, for the quarter, makes for an increase of 60,000 returning users – so it’s actually quite significant.

Since redesign we’ve seen the proportion of Australian use increase from 42% to 51%. Given that Trove target audience is Australians this marks a significant increase in delivering to our key audience. The Program has so far delivered on its aims to refresh the brand, improve user experience and increase user discovery of small and large collections by all Australians, anywhere. Never has this service been timelier than now.