Museo Opera Duomo Firenze: the story of a mobile app at Museum Digital Transformation 2017

The talk below runs about 30 minutes and comes with slides that help clarify some of the key points.

Since the text is long, the slides make a handy summary!

The new museum, which I don't think needs much of an introduction here, is a 6,000 sq m space that was restored and fitted out in truly record time. It took less than 3 years to open it to the public, and some technology decisions played their part in that record.

For a number of reasons, by August 2015, just 3 months before opening, the Opera still had no clear idea of what to do about what is now a key part of the museum experience: the app that has by now replaced the old audio guide. It's a situation almost every museum in the world finds itself in.

Today I'm here to tell you about this story with a “happy ending,” the result of a series of bold choices and of real chemistry between vendors, curators and managers, which we hope can help and inspire others.

Among the distinguished speakers at this conference, my name means little or nothing, and honestly I feel a bit like a fish out of water! It is true, though, that I led the development team behind the case I'm about to present.

I'll take just a minute to introduce myself, which I think I owe to those who came all this way to hear this case study. I've been in IT for 20 years now. In the early 2000s I spent a long and enjoyable stretch as Art Director at an IT company focused on culture and innovation, where, back when the internet was still little more than a pipe dream, I was able to research and experiment with multimedia in museums, building both installations and content management and control systems. Later, my artistic side made room for the joy of writing software, which brought me to Cantiere Creativo, an agency specializing in web solutions that has, in a way, inherited my passion for and experience in the cultural sector. For four years we've worked closely with the Opera del Duomo, and in Florence we're carrying out major projects for both Santa Maria Novella and Le Gallerie degli Uffizi, whose new website will launch in the coming months. You can imagine how thrilling these names are for proud Florentines like us!

To better understand the context in which this project came to life, I think it's important to give an overview of the relationship that developed between the Opera del Duomo and Cantiere Creativo, which is, to all intents and purposes, an organization external to the institution.

Ours is more of a partnership than the classic client-vendor relationship, thanks to a deep trust that has been built and strengthened over time. I'd like to take this public opportunity to give special thanks both to former President Franco Lucchesi, a far-sighted leader who brought outstanding results to the Opera, and to Alice Filipponi, head of marketing and communications, who, besides organizing this event, personally oversees the projects involving our agency.

That trust, as I was saying, rests on a few factors, both qualitative and methodological:

• No contractual constraints or lock-in: everything we do, from source code to design files to the registration of small services, is done in the name and on behalf of the Opera, and nothing remains owned by or registered to Cantiere.
• Full transparency: thanks to detailed, at times obsessive, reporting of what we do day by day, hour by hour, on every single project
• Diligent respect for agreed timelines and costs: a key element in building trust, even more so when it comes with average savings of 35% on almost all the budgets for the projects we were assigned
• Full willingness to change course along the way: this allowed us to refine products based on real needs rather than on overconfidence in the projects' initial ideas
• Solid products: over 4 years, the products we built recorded 99.5% uptime. This meant avoiding disruptions of any kind for both users and the institution
• One team: from day one, Alice and Cantiere's developers and designers were considered full members of a single team, which let us streamline processes and share mistakes, responsibilities and successes too.

On this basis, together we completed more than 10 projects. The first phase, perhaps the most delicate and important of all, was to give back to the Opera what belonged to the Opera, providing all the support needed to recover assets that over the years had been lost to red tape or to the bad habits of industry vendors.

Then the hands-on phase of website development began. We started with a very simple landing page announcing the transformation of the museum complex into a “museum square,” thanks to a single ticket for access.

Besides the institutional website, a must for an institution like this, for the unified museum we looked for a communication style closer to the quality standards of the websites of major international museums. The landing page of the Grande Museo del Duomo, with its new brand, then gradually grew into a complete, comprehensive website in 11 languages, designed mainly for visitors and tourists.

One of the first problems we tackled in 2013 was online reputation: our analysis clearly showed that, with no social media marketing strategy in place until then, the media gave more prominence to disputes with the city administration than to visitors' enthusiastic feedback. That's how the idea of an Instagram contest came about, along with a platform for publishing and rating the photos users tagged.

The numbers went through the roof: we went from about three thousand mentions with a vaguely negative tone to forty thousand, all positive, for a total of one hundred twenty thousand shares and user interactions. The initiative later led to a photo exhibition, so popular and well attended that it was extended by a month, as well as a limited-edition photo book featuring the best shots.

On the road to the new museum's opening, we expanded the website to document the various phases of the project, up to publishing a complete section dedicated to this new area.

Meanwhile, again for the new museum, we delivered the first project on the collection of Choir Books (Codici Corali). In the room called the “Cappella Musicale,” we installed a touch screen that lets visitors explore, read, listen to and virtually leaf through the three volumes on display in the cases, which are swapped every three months. After that, we started rewriting the entire Choir Books management system, which gives curators detailed control over the information on more than twenty thousand items, including images, illuminations and text transcriptions.

This application exposes APIs that let us centrally manage every place where the information is displayed: the screen in the museum, the website, the app and a second touch screen, to be installed in the coming months, that will make the entire catalog browsable.

Last year, we carried out a UX and user interface study for the external ticketing platform. Once the new interface was built and rolled out, online ticket sales immediately rose by about 40%. Then, over the winter, we launched the project Winter Sale, a brand-new online ticket office wholly owned by the Opera. This experiment exceeded expectations in terms of stability, simplicity and ease of use.

The latest project, not a technology one this time but a branding one, was creating the brand, design and merchandise for this conference, which we sincerely hope will be held again in the coming years, because we find it truly important for the sector and for a city with more than 70 museums.

Which brings us at last to the real topic: the museum app, which in the development timeline came along in August 2015.

The curators' requirements were very clear:

• They wanted to completely replace the old audio guide system
• They wanted to give visitors more information about what they were seeing, avoiding crowds in front of the artwork labels
• They wanted to help future visitors understand the experience better before even arriving at the museum, starting at home
• They wanted to centralize information about the artworks, to rule out errors when content was duplicated.

To achieve this, there were two major critical issues:

• the first was time: just over 3 months to make decisions, prepare the material and get everything up and running

• the second was choosing which product to invest in, and here again there were two scenarios:

  • the first was developing a proprietary product: while costs and timing were a major unknown, it offered advantages in terms of control over the product and room to evolve over time
  • the second was adopting an off-the-shelf system, a service already on the market, where “time and costs” were, at least at the sales stage, certain and well defined, but where there was no control whatsoever over the app.

With so little time, the second option seemed the simplest and most obvious. The solutions on the market had their strengths and their limits. In almost every case, they came with long-term contractual commitments, sometimes with a percentage taken from the admission ticket. While those two points might have been acceptable on balance, not owning the content and being unable to centrally manage the data from the official ministry artwork records made this option unappealing.

The alternative, building an application in-house, confronted us with another set of challenges in which time was a crucial variable.

1. Spaces still under construction and not yet accessible for visits, site inspections or realistic tests; in many cases we went on gut feeling until just a few days before opening.
2. The goal was to communicate with visitors based on their position inside the museum, giving them information about what they were actually looking at, at that moment, in that exact spot in the room. Now, the space is vast, split across 3 floors that merge back into a single volume near the great façade. In some areas many artworks sit close together; in others there are just a few standout pieces far apart. Finding a balance and choosing the best technology to physically locate a device inside the museum was not easy.
3. Centralizing the data for each artwork according to ministry guidelines, and avoiding transcription errors spreading across websites, applications, books, catalogs and labels, required work on the cataloging system created by the University of Pisa. Under our supervision, they had to extend it, modify it and build APIs from scratch to retrieve its information in real time, including photos of the artworks.
4. For the app, we expected that, at least initially, most visitors would only become aware of it upon arriving at the museum. That meant guaranteeing Wi-Fi connectivity for a large number of devices and a large volume of downloads. At that point, nothing about the infrastructure was certain yet.
5. On top of all this, it was hard for developers to choose the right technology to cover as many mobile devices as possible across iOS, Android and Windows Phone.

It would have been unprofessional of Cantiere to give advice or reassurances amid so much uncertainty, even though our gut told us we had the skills to build what was wanted.

As a result, it was hard for the Opera to make a decision based on premises full of doubts, and this is exactly where the administration bravely agreed to invest in a period of analysis and testing to assess whether the project was feasible. First, a budget was allocated for the entire app development, with part of it set aside for the analysis, which lasted a month; since neither the scope of the work nor even its feasibility was known, everything was highly hypothetical.

So we started studying the technologies and testing both in our studio and in the accessible rooms of the museum, and within a couple of weeks we realized that some things, despite our doubts, were doable, while others would remain a big unknown until opening day.

For geolocation, GPS, though desirable in terms of accuracy, was not an option given the building's architecture. Wi-Fi would have been useful for triangulation and precise user positioning, but here too we lacked the raw material for testing: the network itself.

That left small Bluetooth-based transmitters called beacons. The issue was that there were a lot of points to cover, almost 350, and frankly we had no hands-on experience with these devices. The first tests failed because of overlapping signals and misread signals. As the days went by we gradually got the hang of them, until we adopted this technology without much hesitation. In any case, given the overall conditions, the trade-off seemed more than acceptable.

Building a realistic prototype that simulated the interface allowed us to achieve two important things:

• A usability study, imagining what the ideal interaction would be for users moving between rooms and floors, and what the biggest usability hurdles might be: one that comes to mind above all is having to turn on Bluetooth on every device. This approach brought to light both the technical and design issues and their possible solutions.
• Second, the prototype let us produce a video for the Duomo's board, with a demo of the app showing them the kind of product we had in mind and making the end result unmistakable, even if not 100% accurate at that stage.

The University of Pisa was very willing to modify its artwork records management system; its technicians extended parts of the database and built APIs to fit our needs. Even so, we couldn't get what we really needed: some artworks are permanent, but others are subject to loans or restoration. Also, archive images are often not of the quality you'd want for a public-facing product. We realized we needed an intermediate application that could pull official content from the catalog, add museum-specific information and pass it to the mobile app via API. Among its tasks was the delicate job of linking artworks to the beacons that represent them and placing them virtually on the map, room by room, as intuitively and quickly as possible for the museum's curators.

Then came the long debate over the type of application, a perhaps boring but crucial topic. Native development, where apps are built specifically for the phone's operating system, certainly allows smoother interaction and more control over the device. Hardware controls for Bluetooth come to mind, or access to the camera or geolocation. But that route risked tripling both development time and costs across very different platforms. The alternative was an app built with web technologies, running inside a browser without worrying too much about the phone's operating system. Although the web isn't the ideal environment for this kind of product, luckily at that time we were starting to put our trust in ReactJS, a framework born inside Facebook that enables very smooth, fully native-style interactions. Wrapping the browser in a cross-platform framework like PhoneGap then let us interact easily with the device and publish dedicated apps on the Apple Store and Google Play.

It would have been nice, really very nice, to test before opening, perhaps with dozens and dozens of devices connected at once, and watch the system under full load. But, as I said, that wasn't possible. The real problem, which dragged on over time, was Wi-Fi connectivity, and it was far from trivial. Connectivity had to cover the entire square: we're talking about roughly forty thousand sq m, with monuments over eighty meters tall, not to mention the interiors. On top of that, access had to be limited to visitors only, so only after picking up a ticket, and not open to the thousands of people who pass through, linger and work in the square.

In short, it would take a whole talk just to tell this incredibly difficult story, which, luckily, wasn't in our hands! Yet in a way we depended on it, and the whole system could only be considered working right before the official opening. A real leap in the dark that cost us a few weeks of sleep!

The app was designed to be very lightweight in terms of megabytes, to allow many fast simultaneous downloads on arrival at the museum. As a result, artwork data was downloaded in real time during the visit. At the time it seemed a sensible choice, but it caused quite a few problems: users struggled to connect and log in to the Wi-Fi, the network was slow and needed fine-tuning, and so on. By monitoring user behavior during the museum experience and listening to feedback, which, alas, included several negative ratings on the stores (and we know how hard those are to recover, being averages!), we let the data convince us that the solution was to make the app usable offline as well, inviting users to download it at home when buying their tickets online. This investment greatly improved the experience, drastically cutting complaints both online and at the desk.

Observing behavior, keeping an ear to the ground, listening to user feedback and, above all, collecting strictly anonymous but extremely valuable data to guide future development was the first step chosen for investment. The upcoming release will let curators monitor a huge amount of data on visitor behavior inside the museum. For example, we'll know average visit times, which rooms are most visited, or how long visitors spend admiring a particular artwork. We'll be able to see which artworks spark the most interest by looking at which multimedia and text content is viewed most during the visit.

Keep in mind that what we built is, to all intents and purposes, a Minimum Viable Product: the minimum product needed to go to market and validate the initial idea, looking for confirmation or dispelling existing doubts. Although it's not a complete, polished product, and it has some gaps and even a low rating, it meets all the curators' requirements, and they ultimately judged the decision to develop a proprietary product a wise one: that satisfaction was the main goal. The real surprise, the icing on the cake, was the final accounting, which revealed a total spend well below expectations.

At the end of the feasibility phase, we realized we were further ahead with research and development than we could have imagined in our wildest dreams, and we had the app finished and published before even 65% of the budget had been spent. Only with the updates and maintenance over the following 16 months did we reach 90% of the initial budget allocated in 2015.

From a project management standpoint, this can be considered a real success.

Now that we have a solid foundation, what will we do with this app? We have some ideas: some are on the starting line, others need more exploration. Here's a quick overview that I think helps show how digital is truly transforming this field.

The app will certainly be expanded beyond the museum to cover all the monuments of the complex and the artworks inside them. Visitors will be able to use it in the bell tower, the Baptistery, the Cathedral and so on.

This expansion, combined with the data monitoring system and device geolocation, will let the Opera's staff, at peak crowd times, notify visitors who may have been waiting in line for hours and suggest an alternative route that is less busy at that moment, helping redistribute the flow in a way that works better for everyone.

We're already working on extending the app with features for deaf users, adding navigation modes that will provide purpose-built multimedia content.

The content of this first version of the artwork records will be expanded with additional texts, galleries, audio and video. At first this will cover only the main artworks; then, based on the behavior observed through monitoring, it will extend to the artworks visitors find most interesting.

Visitors will be able to photograph artworks without leaving the app and share the photos on their favorite social channels. Also on the social side, visitors who wish to will be able to see content or comments posted by others at the same time, interact with them virtually or, why not, in person, since they're in the same place. Who knows, maybe we'll hear someone say “It was the museum that brought us together…!”

Integrating the online shop, both for users physically at the museum and for those using the app at home. The app will offer suggestions related to the artworks, such as books or merchandise to pick up at the bookshop on the way out or to have shipped home. Ticket purchasing will also be integrated, so visitors can skip the ticket office.

Finally, the tech buzzword of the moment: the chatbot, how could we leave it out!? Interaction with bots has already started on all our phones, and it won't be long before we're at home in the kitchen talking to the fridge, and maybe it will even answer back! A bot smart enough to answer questions and satisfy curiosity about an artwork, a monument or an artist, providing answers or multimedia content tailored to the user's request, could open the door to a new kind of museum experience.

There are agencies that know how to develop these technologies, as we do, but on your own you don't get very far. You need to build synergy and collaboration with those who have real expertise and authority over the content, and work side by side with them.

For us, the experience with the Opera del Duomo confirmed this fine theory.

Thank you.