<p>3D visualisations –including 3D scans and 3D reconstructions–designed as part of larger archaeology, history or cultural heritage projects are commonly shared with the public through online platforms that were not necessarily designed to host heritage representations and often fail to contextualize them. This paper seeks to evaluate whether five online platforms commonly used today to share 3D visualisations of heritage (Google Arts & Culture, CyArk, 3DHOP, Sketchfab and game engines) offer features that facilitate their scientific rigour and community participation, based on guidelines from International Council on Monuments and Sites(ICOMOS)and United Nations Educational, Scientific, and Cultural Organization(UNESCO). The author starts by summarizing recommendations from 32 international guidelines that are relevant to the 3D visualization of heritage, condensing them into nine key criteria: multi-disciplinary teams, objective-driven methodology and tools, careful documentation, type of reconstruction and level of certainty, authenticity, alternative hypotheses, multiple historical periods, respectful use of the heritage, and community engagement. The author proceeds to review the platforms above comparing their features with these nine recommendations and concludes that, while there are currently available features that could help to elevate the scientific rigour of the 3D visualisations and their contextualization to the public, they are not mandatory and are seldom used. The paper finishes with a recommendation for an information package to support3D visualisations of heritage on public online platforms.</p><p><strong>Highlights:</strong></p><ul><li><p>Online platforms for the 3D visualization of heritage fail to disclose what type of reconstruction it is and its level of certainty, struggling to balance community engagement vs scientific rigour of their contents.</p></li><li><p>ICOMOS and UNESCO recommendations regarding heritage are loosely followed on the reviewed platforms, and supporting documentation is often lacking.</p></li><li><p>Scientific rigour on these platforms could be elevated with supporting textual fields to disclose further information about each visualisation.</p></li></ul>