SMK has already made the choice
The National Gallery of Denmark (Statens Museum for Kunst, SMK) is building a digital visitor guide that opens in the browser, is scanned with a QR code and runs on the visitor's own phone. There is no app. The project is called SMK Link, and the museum plans to launch it in 2027.
Denmark's best-resourced art museum has looked at how visitors actually use digital guides, and has dropped the model many museums still pay for: a native app that the visitor has to find, download and understand before the interpretation can begin.
Jonas Heide Smith, head of SMK's digital unit, writes it openly on the museum's blog. SMK Link is a browser-based mobile guide to the museum and the art on the walls. The interface is built on QR codes and the visitor's own device. The choice reflects, he writes, the museum's scepticism towards native apps.
If you are sitting with an app development, a renewal of an old hardware guide, or a wish for multilingual interpretation, this is the right place to start. Read what SMK and the Danish museums that already run apps themselves say about what visitors do.
What SMK Link actually is
SMK describes two situations.
In the first, the visitor comes into the foyer looking for a tool that can help them find their way around. Here the guide should show exhibitions, the audio guides that already exist, and an interactive map so that the size and layout of the rooms make sense.
In the second, the visitor moves freely through the galleries and stops at a work she wants to know more about, or simply to read about in her own language. By the QR code sit icons that tell her what the scan gives: audio, sign language, video, fuller descriptions.
Smith calls the second situation the most interesting, because it requires the least preparation and the least prior knowledge. The visitor does not need to have planned a digital trail before she walks in. She scans when curiosity hits.
The scan is also a signal to the museum. It tells the museum that the visitor is interested in that particular work. SMK will use it to compile a personal list, to suggest the next stop, and for wayfinding: where am I, where have I been, have I missed something, are we done soon.
An app can do all of that too. The difference is that SMK refuses to make download the ticket of admission.
In a later post, from June 2026, Smith writes that SMK is building the guide from scratch as a browser-based solution. The 1,500 planned, expanded artwork texts are written by people in Danish and translated by people into English. Machine translation is to cover French, Italian, German and Swedish. Text-to-speech is to run in every language. The purpose is accessibility: visitors who only speak Italian, visitors with a visual impairment, visitors with dyslexia, and visitors who would rather listen because their eyes are occupied by the work.
Then comes the sentence that locks the timeline: drop by and hold us to the promises when we launch SMK Link in 2027.
The evidence behind the choice
SMK points to findings from Rosa Sala Rose, chief executive of Nubart: very few visitors use museum apps. Smith adds his own observation. He rarely uses museum apps himself. He expects hassle with installation and more hassle with an arbitrary interface. In an unfamiliar museum he is already overwhelmed. And he rarely sees a clearly stated reason to install that outweighs it.
Nubart itself supplies QR-based guides. That should be said first. 2.47 per cent is not an independent official report, and the figure does not appear in SMK's own texts. It is Nubart's estimate, based on 175 official audio-guide apps on Google Play in Europe and the USA. Third-party apps are left out. Platform apps such as Bloomberg Connects are also left out, because downloads cannot be attributed to the individual museum. For each app they have estimated annual downloads from Google's public ranges, adjusted for the iOS share and divided by the museum's annual visitor numbers.
Nubart itself stresses that the figure counts installations, not use. It includes people who installed the app and deleted it again, and people who never opened it. Actual use is therefore likely to sit lower. Read 2.47 per cent as a supplier figure with caveats. SMK has found the overall conclusion credible enough to drop native apps.
According to Nubart, the exception is Museum Barberini in Potsdam at 20.76 per cent, built with resources most museums never get. Even there the app only reaches around every fifth visitor. Barberini is the ceiling, Nubart writes, not a realistic target for a museum with an ordinary budget.
SMK quotes the sentence that has since become the sharpest formulation of the problem:
Museums are experts at removing obstacles to cultural experiences. But in the digital realm, many have unintentionally put one of the biggest obstacles in the way: requiring a download.
That is the evidence SMK is acting on. Apps can be rich in features. They still lose most visitors before the interpretation has begun.
The Danish admission: download from home
In Denmark you can see the same weakness without opening an international white paper. It sits in the text the museums themselves write to visitors.
The Viking Ship Museum (Vikingeskibsmuseet) plays its audio stories in the Useeum app. The app is downloaded free from the App Store or Google Play. The visitor should bring a smartphone and preferably headphones. And then comes the sentence that says more than the entire app store feature list:
Scan the code and download the app from home, so that you are ready to solve the mystery and hear stories as soon as you arrive at the museum.
Read it as it stands. The museum knows that the visitor has to be ready before she walks in. The solution is to move the installation out of the visit and onto the visitor's preparation. It works for the visitor who plans the trip three days in advance, has space on the phone and remembers to do it. It works poorly for the international first-time visitor standing in the foyer with a suitcase, children and twenty minutes.
Several Danish museums recommend downloading the app from home because the wi-fi is unstable. That is an admission of the app model's weakness. The app requires both an action the visitor would rather avoid, and a network the museum often cannot guarantee, precisely when the action has to happen.
The interpretation behind the app can be strong. The Viking Ship Museum has trails for children, adults and experts, in Danish, English, German, French and Spanish. The problem is access. The content sits behind an installation the museum itself warns against carrying out on site.
If you want an overview of the four main types (hardware, app, QR, AI narration) and the typical prices, it is in our guide to choosing an audio guide. That article is about product types. This one is about what visitors actually do once you have chosen.
Where the app loses during the visit itself
A native app can be excellent once it is running. Offline content, push notifications, maps, gamification. The problem is the steps before.
The visitor has to discover that the app exists. She has to find the right one in an app store full of museum names, city guides and old versions. She needs an Apple ID or a Google account, storage space and patience for permissions. After that she has to find her museum inside the app, if it is a platform such as Useeum. Only then can she listen.
Each step sorts someone out. Nubart's method cannot count how many gave up along the way. It can only count those who got far enough to install. Even if one takes 2.47 per cent at face value, it is a ceiling for installation, not a measure of listening.
Smith's requirements for SMK Link are the reverse of that chain. No pop-ups. No newsletter sign-up. No waiting time. Very little navigation. The guide should feel so natural that the visitor thinks it was easy to build. It will not be, he writes. It just has to be experienced that way.
That is why the browser wins on the evidence, even when the app wins on the feature list. The browser removes the step the evidence says is fatal: download.
Browser and QR are the same architecture
A browser guide on the visitor's phone means that the visitor uses what she already has in her pocket. The museum makes the content available, not a new app and not a device handed out.
The QR code is the entrance. iOS and Android have a QR reader in the camera. The visitor points, and the page opens. That part is familiar. The practical set-up (size, placement, brand colour, testing) we have described in the guide to QR codes in museums. The point in the evidence is a different one: the QR code is only better than the app if it lands in the browser without a detour via the App Store.
SMK's icons by the code are the detail many museums forget. The visitor should be able to see what she gets before she scans. Audio, sign language, video, longer text. Without that, the code is a promise without content, and promises without content you scan once.
Wi-fi does not disappear as a problem. A browser guide still has to fetch a page. The difference is that the visitor is spared completing an installation under the same poor conditions. SMK chooses that model deliberately, well aware that the browser has limitations. Those limitations weigh less than the download requirement.
When you assess a supplier, there is therefore one question that separates the architectures: does the scan land in the browser, or does the code send the visitor on to the App Store? If the answer is the latter, you have been given a QR code with the app model's barrier built in.
What it means for smaller museums
SMK is not a typical Danish museum. The museum has a digital unit that has worked with an open collection for years. The dataset of around 45,000 works has been used by AI researchers globally. Lisbet Tarp, associate professor at Aarhus University (Aarhus Universitet), calls SMK a pioneer in AI interpretation. Jonas Heide Smith can allow himself to build a guide from scratch and say 2027 out loud.
Most state-recognised museums do not have that capacity. They have an interpretation lead, a communications officer and an IT budget that is already spent. For them, SMK's choice is still the most important signal in 2026. If Denmark's national gallery concludes that native apps are the wrong visitor barrier to build, it is hard to argue that a museum with a fraction of the budget should pay for the same barrier and then ask visitors to download from home.
Denmark's best-resourced art museum has chosen the same architecture as a browser and QR guide: the visitor's phone, no app, scanning at the work. They are building it themselves. They launch only in a year. If you wait for SMK Link to become a finished off-the-shelf product you can copy, you are waiting for an internal development project that was never designed for you.
Museum Helsingør shows another Danish path, still with the visitor's own device as the starting point. Together with Digital Impact the museum has launched "To Prompt", described as probably the world's first AI-based history guide to the city. The model is trained on information from around 18,000 articles and archival records. Project manager Sebastian Gyrst-Longsig Christensen points to the collection that never makes it onto the wall: 1,200 linear metres of archival records and almost 40,000 objects in store. AI makes it possible to put them into play.
That is a different product from SMK Link. It is the same direction. Meet the visitor where she already is, with the device she already has, without asking her to install the museum's software first.
You do not need to wait until 2027 to build it yourself
SMK Link validates the category. It does not prove that every museum should hire a digital team and write 1,500 artwork texts by hand.
If you want the visitor experience SMK describes (scan, open in the browser, listen in your own language, no app), it already exists. At museum-guide.com we make multilingual audio guides that open directly in the browser via a QR code on the visitor's phone. You supply the texts. We help with translation, narration and the page the scan lands on.
You do not need to wait until 2027. You do not need to build it yourself.
The evidence SMK has put forward is there to act on now: drop the download requirement, meet the visitor in the camera, and let the interpretation begin while she is still standing in front of the work.
Sources
- Jonas Heide Smith, SMK blog: Building a digital museum guide that guests may actually want
- Jonas Heide Smith, SMK blog: AI in our new visitor guide? Very tightly curated
- SMK: Artificial intelligence (AI) at SMK
- Rosa Sala, Nubart: Museum Audio Guide Apps: Only 2.47% of Visitors Download Them
- Viking Ship Museum: Audio stories - Audio guide
- Dagens AI: SMK is world-leading in the use of AI for museum interpretation
- Kulturmonitor: In Helsingør a city museum uses AI to reach more visitors
About the author: Thøger Elung-Jensen is co-founder of Audioguiden and works to help museums use AI to take their interpretation to more people.