Analytics for Crowdsourcing Art & Design

For over a decade, Minted's community of artists and its crowdsourcing platform have delivered perpetually fresh, high-quality art and design into the world. We continue to expand our rich, granular data on the interactions, the designs, and the votes of the community. Minted Labs uses that data to understand and further the development of artists, the community and - above all - great art and design.

Current Research Focus


There is a common misperception that the crowdsourcing of designed products is the simple act of sending product images to the crowd, and then soliciting and tallying votes. In reality, the commercial success of a collection of products is determined by how the voting process is structured, how the voters are solicited, and how one identifies and weights those voters who are most predictive of sales. We have spent over a decade developing processes and algorithms to do that, and continue to make this a major focus of our research.


Preserving the vigor of our community and the purity of our crowdsourcing process is of utmost importance to us, and our historical voting and interaction data enables that. We are currently pursuing the following questions: What drives people to vote and how can we encourage it more broadly? How can we create voting processes that remove unintentional bias toward certain design styles, artists, colors, or photos? How do we focus our process so that it delivers on our mission of identifying the world's best art and design?


The instrumentation of our platform is extensive, and we have been storing granular data about how artists on the Minted platform engage, interact with each other, and develop professionally. We study the behaviors that lead to successful outcomes for artists, and then ask ourselves how we can create experiences, processes, and tools that foster those behaviors. We are perpetually curious about how artists, both in the Minted community and in the world in general, develop their talent and their commercial success.


We have a massive archive of designs but have only scratched the surface in terms of analyzing long-term trends. We have a robust, low-cost, scalable approach for attributing and attaching metadata to our archive, and can cheaply create learning datasets. The next steps are to design a generalized taxonomy that lets us see trends across product types, and the development of zero-cost, machine-based image recognition processes that allow us to massively scale the volume and breadth of our attribution.

Driving great Art & Artist Development

Academic Partnerships

We are looking for academic partners who share our interests in the application of analytics and technology to the understanding of the development of great art and artists. If you are interested, please get in touch at .