How museums can adapt their content to short-video formats

For a growing share of your potential audience, the first encounter with your museum is a vertical video they did not go looking for. It arrives between a recipe and a dog, it lasts forty seconds, and it decides whether they ever think about you again.

Vertical video is now the default unit of social media. TikTok, Instagram, YouTube, Facebook, LinkedIn, Threads and X all push it, and all of them reward organisations that make it natively rather than repurposing whatever the marketing team already had. The question is no longer whether to make short video. It is how to make it well without a production budget you don’t have.

The landscape in 2026

TikTok is still the strongest engine for reaching people who have never heard of you, because its recommendations are driven by interest rather than follower count. A museum with 800 followers can still put a video in front of 200,000 people. The caveat for 2026 is structural: TikTok’s US operation is now run by a majority-American joint venture, with the US algorithm retrained separately. In practice this means US and international audiences may increasingly see different things, so if your audience spans both, don’t assume one video will behave the same way in both markets.

Instagram Reels has stopped being a bolt-on feature and become Instagram’s main discovery mechanism. Reels are tested on non-followers first, which means quality matters more than loyalty. Saves and shares carry more weight than likes, so someone sending your video to a friend with “we should go” is worth more than a tap on the heart. That save-to-view ratio is the single most useful number for a museum to watch.

YouTube Shorts has quietly become the most useful platform for museums with genuine expertise. Shorts now run up to three minutes, monetisation has matured, and, most importantly, Shorts sit inside YouTube’s search index. A two-minute clip explaining how a conservator removes varnish will keep surfacing for years. TikTok and Instagram content largely doesn’t.

Facebook Reels is the one most cultural organisations dismiss and shouldn’t. If your visitor base skews over 40, as most museums’ does, this is where they are. The same clip that underperforms on TikTok can do real work here, and Facebook remains the platform most likely to drive an actual ticket purchase.

LinkedIn video is worth a thought for a different audience entirely: funders, trustees, sector peers, potential recruits. Short video about how the museum works, rather than what’s in it, performs unusually well there.

The practical implication is that a single-platform strategy is now the risky option. Algorithms shift, policies change, and audiences fragment. Most organisations seeing growth are publishing the same core material to three or more places, with light adjustments to each.

Length has changed, and so has the metric

The old advice was “keep it under 15 seconds.” That’s out of date. All the major platforms have expanded their limits, and their algorithms now reward longer watch sessions. Instagram Reels can technically run far longer than three minutes, though the discovery algorithm stops recommending anything past roughly that mark to non-followers. YouTube Shorts caps at three minutes. TikTok allows ten.

But the number that matters isn’t length, it’s completion. A 20-second video watched to the end will beat a 90-second video abandoned at 15 seconds, every time. The right length is the shortest one that fully answers the question you raised in the first three seconds. Padding a video to hit a target does active harm.

For most museum content, 30 to 60 seconds is the sensible default. Go longer only when the material genuinely earns it: a conservation process, an object with a real story, an expert who is actually good on camera.

Short video is now a search engine

This is the biggest strategic shift of the past few years, and the one most museums are still missing.

A significant proportion of younger audiences now search on TikTok and YouTube rather than Google. They type “things to do in Rotterdam when it rains,” “is the Rijksmuseum worth it,” “free museums Manchester,” “what to see in one hour at the British Museum.” Platforms index your spoken words, your on-screen text and your captions.

That gives museums an enormous, largely untapped opportunity: make videos that answer the questions people are already asking. Practical, unglamorous, evergreen content will outperform another beauty shot of your atrium: how long you need, what to skip, where to eat, whether it’s suitable for a four-year-old, what’s actually free. It also keeps working long after the trend it might have chased has died.

Say the museum’s name out loud in the video. Put the city in the on-screen text. Write captions as sentences a person might search, not as strings of hashtags.

Your unfair advantage is that you are real

Generative video flooded social feeds through 2025 and the backlash has been sharp. Consumer preference for AI-generated creator content has fallen steeply, platforms have started demoting synthetic-looking material, and OpenAI’s AI-video social app Sora, briefly the most-downloaded app in the US, was shut down in spring 2026 after interest collapsed.

The lesson for museums is not “avoid AI.” It’s that in a feed full of plausible nonsense, verifiable reality is worth more than it used to be. You have objects that exist, spaces people can walk into, and staff who genuinely know things. Slight imperfection now reads as proof of life: a curator stumbling over a word, a lighting rig visible in shot, a phone camera rather than a cinema camera. Don’t polish that out.

Where AI does earn its place is production, not creation. Auto-generated captions you then correct, clipping a recorded lecture into six usable segments, drafting caption variants, transcribing footage so you can find the good bits. If you do publish anything synthetic, whether a reconstruction, an animation or a generated voice, label it. Platform terms mostly require it, EU transparency rules increasingly do, and museums trade on trust in a way that brands don’t. It is not worth the risk.

Three things that will quietly cost you

Music rights. Institutional accounts do not have the same access to platform sound libraries as personal accounts do, and museums are, in most jurisdictions, commercial entities for this purpose. Using a trending track you’re not licensed for can get a video muted, removed, or the account restricted. Check what your business account is actually entitled to use, and build a bank of cleared audio.

Watermarks. Reposting a TikTok export, logo and all, to Reels still suppresses reach. Export clean, add sound natively on each platform.

Captions. Burn them in rather than relying on auto-generated overlays. Most people watch with sound off, auto-captions mangle names and specialist terms, and for a sector with an access remit, getting this wrong is a bad look as well as a lost audience.

Eight formats worth trying

The best approach is still to experiment and see what your particular audience responds to. These are the ones that consistently work for cultural organisations:

1. One object, one story. The oldest format and still the best. A single thing, one surprising fact, sixty seconds.
2. Answer a real question from the comments. Free content ideas, and it signals you’re listening.
3. Conservation and behind the scenes. People are fascinated by hands doing careful work. Store rooms, mount-making, packing, installation.
4. The practical visit guide. How long you need, where to start, what’s free, best time to avoid crowds. Search-friendly and genuinely useful.
5. A named expert, recurring. Series beat one-offs. A curator or educator who appears regularly builds an audience in a way a rotating cast never does.
6. Myth-busting. Every collection has one. Correct it in forty seconds.
7. Front-of-house voices. Visitor services staff know which questions get asked forty times a day. That’s a content list.
8. Creator collaborations and user-generated content. The Rijksmuseum’s approach of inviting creators into the building and letting them make their own work reaches audiences the museum could not reach directly, and costs far fewer staff hours than producing everything in-house.

Finally: do less, but properly

The most common mistake in 2026 isn’t bad video. It’s scattered video: three platforms, no pattern, whatever anyone had time for that week.

Pick one repeatable format. Film four or five of them in a single session. Post consistently for three months. Watch saves and shares rather than views, and watch whether any of it shows up in your visitor numbers. Then decide what to do next based on what your audience actually did, rather than on what a platform’s latest announcement said they would.