Most museums know the public funding squeeze all too well. Attendance swings with the seasons, the economy, and the weather. Costs are increasing. Every year the task of just staying open gets harder.
The typical answers are to simply get more people through the door, or to ask donors for more money. But museums already hold within them a valuable asset that doesn’t need either: their digitised collections. High-quality images, metadata, 3D scans, footage, audio recordings, and other archival and scholarly material have all already been created, but are being seen by a fraction of the people who would find it useful.
Licensing offers the chance to take that material and not only give it new life, but turn it into recurring revenue. It doesn’t require the time and financial investment of new exhibitions, more gallery space, or bigger visitor numbers. It just requires knowing what you can license, and having a practical way to do it.
What “Licensing” Actually Means
A license is giving permission to do something with an asset on stated terms. Ownership stays with you as the museum, and you dictate the terms of the license. Everything from usage rights to time periods to use cases is covered in the agreement, giving the museum the power to decide exactly what a partner can and can’t do. It’s important to remember that a museum isn’t simply selling its entire collection, but granting a defined right.
What Museums Can License
What museums can license is usually wider than most would think:
The objects and artworks - while older artworks are often public domain (more on that later), modern and contemporary works are usually still under copyright by either the creator or their estate. The museum cannot license rights to a work it doesn’t hold.
Images or digital replications - this is where it gets trickier. If a museum has created a faithful reproduction of an out-of-copyright image - for example, by creating a high-quality scan - then it typically doesn’t own the copyright for this image. If an image has additional creative choices - for example in framing or composition - then that can create a new copyright. Regardless, it can still license the image, though.
Catalog data and metadata - this covers things like descriptions, provenance research, tagging, and more. Often this is more valuable to AI companies than the artefacts themselves, as it’s the data that helps to train their systems.
3D models and photogrammetry - this can be useful for things like video games, education, and Augmented/Virtual Reality applications.
Audio, video, and archival footage - think film, TV, and streaming.
Texts and other documents - things like audio guides, video, essays, and catalogs are also vital to AI companies.
That said, it’s also important to understand what isn’t licensable. This would cover works with unresolved copyright, culturally sensitive materials that a community might not have agreed to share commercially, or items that have come with donor restrictions. It’s important to know what can, and can’t, be licensed before delving further.
Who Is Buying And Why?
Publishers and producers - books, documentaries, films, TV programs, exhibitions, and video game producers need real, authentic materials for their media, and these materials need to be rights-cleared so they can publish without delay.
AI and technology companies - with much of the open internet already scraped, AI and tech companies increasingly need high-quality, rights-cleared, properly-documented data with clear provenance to improve their systems. A lot of this already exists within museums.
Educational platforms and developers - these need high-quality, rich historical and reference material to power their tools, research systems, and curriculum content. Accuracy and trustworthiness are imperative for these.
Brands and manufacturers - companies in areas like fashion, textiles, homeware, consumer goods, and others want things like heritage patterns, designs, and period imagery for products and collaborations. These can, in some instances, be very lucrative.
All these potential clients and potential uses on the same things: high-quality data. But what does that actually mean?
What Does “High Quality” Mean For Licensing?
In terms of things like AI training data, “high-quality” doesn’t just mean high-resolution or well-written. It actually covers qualities like:
Accuracy
Trustworthiness
Structure
Context
Legality
Digestability
And more
Let’s take a painting from the 1800s as an example. Low-quality metadata might be:
Painting of a field, circa 1800s
This is low quality as it lacks the context, depth, and information required for a system to fully understand the item itself. Who created the painting? What style is it in? Where did it come from? These all help an AI system to make more sense of the object in front of it, and make for far more useful metadata. A higher-quality metadata might look like:
“Richmond, Yorkshire”. Landscape painting depicting Richmond, North Yorkshire, created by J.W.M. Turner in 1818. The watercolour painting shows the town in the distance, with the River Swale in the middle ground and a female in the foreground.
This is far more useful to an AI system as it provides the detail, context, and information with which to understand the work.
Why Don’t More Museums Use Licensing?
If the opportunity is so clear, why isn’t every museum already doing it? Some already are, and in ways that don’t require significant outlay.
The V&A, for example, worked with Samsung to license its extensive collection of artworks for use on Samsung’s The Frame television. Started in 2018, the initiative allows consumers to purchase selected artworks and display them on their TVs when not in use, transforming the screen into a digital art display. The deal is still ongoing some nine years later.
Meanwhile, in 2015, the William Morris Gallery partnered with retailer Jigsaw for a clothing and homeware range. The limited-run items yielded the gallery £25,000 for a single pattern.
But while the benefits are clearly there for museums, there are some obstacles that hold them back:
Inconsistent metadata - incomplete, inconsistent, or incorrect metadata can make the process more difficult.
Uncertainty over rights - museums that don’t understand what they can and can’t license often struggle to commit to the licensing process.
Time investment - the licensing process - from finding a partner all the way through to deciding and agreeing terms - can represent a significant time investment with no efficient way to manage it singlehandedly if a museum hasn’t done it before.
Usage worries - despite terms being agreed to by both parties, there can be legitimate worries that commercial uses might undermine or damage the museum’s credibility if data is used in the wrong way.
While none of these concerns are unreasonable, they are solvable.
How Can Museums Negate These?
Metadata enrichment - this is where, by using AI or a more manual system, context, classifications, and relationships are added to raw data assets. Schemas like Dublin Core or LIDO can provide the framework to improve the metadata.
Risk-assessment - consider sorting your collection by rights status before contacting anyone. All works carry a different level of risk, so starting with the lowest-risk group means a museum can commit to licensing with confidence.
Start small - an institution doesn’t have to go all-in with their entire catalog with their first deal. A pilot built around a well-documented set of objects will be easier to manage and can give a team the evidence it needs to expand its offerings.
Set clear guardrails - a museum can be as restrictive or as open with their deals as they wish. Permitted uses, approved rights, and attribution requirements are all within the museum’s control.
How Many Items Does A Museum Need To Start?
There really is no minimum number of items a museum needs to start licensing as it relies more on the buyer, the deal, and what the museum is looking for.
For individual deals like the aforementioned Jigsaw X William Morris partnership, a single pattern was enough. The V&A’s deal with Samsung was for a handful of selected artworks. In this kind of licensing, it’s the strength of individual materials rather than the size of an archive that makes the difference.
When it comes to AI training, though, it’s far more often about scale. Buyers are looking more for bulk data, as both quality and quantity are important for training. That comes with a caveat, of course, in that some topics might only have a limited amount of data, which can make smaller, specialist collections more valuable.
What counts more is how many items are usable, not just how many exist. A collection of 5,000 items with high-quality metadata might be more valuable than a collection of 100,000 objects with missing or partial metadata.
How Do Museums Start?
Starting to license your content starts with understanding what you have:
Audit your digital assets - have a clear idea of what’s in your collection, and the rights status of each one.
Identify high-demand collections - start with materials that are distinctive, well-known, well-documented, and rights-cleared.
Clean up and enrich metadata - better descriptions, clearer relationships and more data helps to improve discoverability and makes collections more valuable.
Set your institution’s boundaries - establish what you do and don’t want to license, who you are (or aren’t) prepared to license to, the use cases, and anything else about how your data can be used.
Choose a platform - while managing your own partnerships is possible, finding a platform with which to partner can take some of the stress off and help to find new potential partners.
Launch - starting with a small pilot, you can test how the process works in practice and how buyers respond to your materials. A limited first run also means less time investment, allowing you to identify any issues or roadblocks before expanding.
What Next?
Licensing won’t replace the traditional revenue generators of admissions, memberships, merchandise sales, or donations. But it can sit alongside them as a contributing revenue stream that doesn’t rely on footfall, the economy, or the weather. It adds no additional gallery cost, no initial outlay for merchandise, and can give your pre-existing collections a new use beyond the walls of your institution.
Buyers are already looking for the data in your collections. The next step is to find out how popular your collections are, and what you could earn from them.
Interested in finding out more about the Creative Content Exchange? Get in touch with us today!