Why I'm worried when museums hand collections to tech platforms
I've seen this pattern firsthand: a regional museum under budget pressure signs an attractive deal with a well-known tech platform promising "greater reach", sophisticated search and analytics, and a streamlined public interface. The headlines sound great. Donors are pleased. Digital visits tick up.
But too often what gets traded away isn't visible on any balance sheet. As someone who watches how institutions, technology and public life intersect, I want to spell out what museums — and the communities they serve — risk losing when collections are outsourced to third-party platforms. This isn't technophobia. It's about stewardship, power and the long view.
What goes missing: the tangible and the intangible
When collections leave local control, losses occur across several dimensions:
Authority over context: Museums don't just hold objects; they hold narratives. Platforms often standardise metadata to fit search algorithms, stripping out nuance about provenance, contested histories, or local meanings.Access governance: Who decides who sees what? Platforms can enforce paywalls, restrict API usage, or prioritise commercial partners — shifting access decisions away from curators and community stakeholders.Community voice: Local people who have co‑created the stories around objects can be sidelined. A platform's schema rarely accommodates oral histories, subaltern taxonomies, or non‑Western classification systems.Long-term preservation: Organisations can become dependent on the platform's financial health and technical choices. If a startup pivots, is acquired, or shuts down, collections can become fragmented or inaccessible.Data ownership and portability: Contracts sometimes give platforms broad rights to use, analyse and monetise collection data. Extracting and migrating that data later can be costly or legally constrained.Curatorial skills erosion: Outsourcing routine digital curation can hollow out in‑house expertise, leaving teams unable to interpret, correct or re‑contextualise their own holdings.Algorithmic bias and visibility: Recommendation engines and ranking algorithms can invisibilise certain objects or stories, privileging what performs well rather than what matters historically or locally.Real examples that should concern us
Not every platform engagement ends badly — partnerships with Wikimedia Commons, Internet Archive or institutions using IIIF are often positive. But there are worrying cases. A regional gallery digitised its entire local textile archive via a commercial platform; years later, search terms for local place names returned inconsistent results because the platform had normalised names to national standards. A community group's oral histories were given generic tags, losing speakers' names and migration histories that mattered to descendants.
These are not just technical mistakes. They're erasures of memory and authority.
Practical questions to ask before signing anything
Before you agree to a platform, demand clear answers to these questions. If the provider hesitates, that’s a red flag.
Who owns the data? Do you retain copyright and moral rights? Can the platform reuse content commercially?What are the exit terms? Is there a clear, tested export process for all files and metadata in open formats?How will metadata be transformed? Will your original fields be preserved, or mapped to the platform’s schema? Can you add custom fields?Who controls access and permissions? Can you set public/private flags, and revoke access if needed?What happens to backups? Are they stored geographically and in multiple formats? Who has the keys?Will your community's taxonomies be respected? Can you include oral history transcripts, non‑Latin scripts, and local place names?What analytics are provided, and who can access them? Are usage metrics shared with you in raw form?Is there a sustainability plan? What happens if the company is sold or ceases operations?Contractual clauses and technical safeguards I recommend
When I read service agreements for cultural institutions I look for a few non‑negotiables. You should too.
Data ownership clause: Explicitly state that your institution retains full ownership and copyright of uploaded assets and metadata.Exportability clause: Require periodic, full exports in open formats (e.g., CSV or JSON for metadata; TIFF/PNG/MP3/WAV for media) at no additional cost.Preservation and redundancy: Stipulate that backups must be stored in at least two independent locations and that you receive checksums and validation reports.Right to audit: Allow audits of how data is used, indexed and monetised, and require transparency on algorithmic ranking where applicable.Localisation and custom fields: Preserve your original metadata fields and allow custom vocabularies and scripts.Termination and escrow: Define clear termination triggers and a digital escrow mechanism to ensure assets return to you if the provider fails.Technology choices that protect stewardship
Where possible, prefer technologies and standards that make future migrations easier and community engagement richer:
IIIF (International Image Interoperability Framework): Enables deep zoom, annotations and interoperability between viewers — it reduces vendor lock‑in for images.Wikimedia and Wikimedia Commons: Good for public outreach and collaborative curation when rights allow; strong on openness but requires care around sensitive materials.Omeka and CollectionSpace: Open‑source collections management systems designed for museums, allowing local hosting and custom metadata.Internet Archive / LOCKSS: Useful for independent backup and long‑term preservation.Open metadata standards: Use Dublin Core, METS/MODS or Schema.org where appropriate to avoid proprietary lock‑in.A practical checklist for regional curators and community activists
Use this as a working checklist when evaluating any proposed outsourcing:
Demand written answers to the questions above before any pilot project.Insist on trial exports — can you actually retrieve your files and metadata in usable form?Keep an offline backup (at least one copy offsite) before any upload.Protect sensitive content: Put in place community consent processes for human subjects, gravesites, and culturally sensitive items.Negotiate analytics access — your data about your audiences is as important as the assets themselves.Train staff in digital stewardship so institutional knowledge remains local.Make partnerships conditional on community governance roles — include local voices in how collections are presented online. | In-house / Open systems | Proprietary platform |
| Control over narrative | High — full curatorial control | Lower — standardised by platform |
| Upfront cost | Variable — may need investment | Lower — often cheaper initially |
| Long-term risk | Lower if open formats used | Higher — vendor dependence |
| Community co‑curation | Easier to implement locally | Depends on platform flexibility |
When outsourcing can still make sense
I'm not arguing for isolation. For many regional museums with tiny budgets, platforms can be transformative: rapid digitisation projects, searchable public catalogues and enhanced discoverability. The critical point is to treat outsourcing as a partnership that must be negotiated, not a one‑way handover.
Insist on protections, insist on community involvement, and insist on technical choices that respect future control. If you can't get those, keep the collection local, or find alternatives like Wikimedia or local university partnerships that prioritise openness.
In a moment when memory and identity are contested, handing collections to a faceless stack of servers without safeguards is not neutrality — it is a choice about who will speak for the past. If you're a curator, trustee or community activist, it's a choice you should make deliberately and with your eyes open.
