Launched in 2008, artnet’s online auctions pioneered the online only model for fine art auctions and quickly grew to gain a global audience of art dealers and collectors who transacted increasingly higher volumes and value of art across 190+ countries.
In its essence, the platform is a heavily managed peer-to-peer marketplace for fine art, where specialists manage relationships between both buyers and sellers throughout each phase of the exchange from listing, bidding, shipping and receiving. Due to the nature of collecting art, there's a much higher demand for trust and authority than other industries, making the processes and workflows involved to run such systems significantly more complex and more high touch than traditional e-commerce.
Because operational costs are so high, it was crucial to automate the more manually intensive workflows, such as the processes of accepting works from sellers that are eligible to become part of a sale. As part of the product team we were tasked to identify ways to make these processes more efficient for all marketplace participants as well our internal teams, especially around our authentication and admission workflows.
The challenge
Manual review of every artwork submitted worked well in the initial “start-up” phase of the business, but it soon represented a major blocker in its growth trajectory. The need to keep overall operations and overhead cost low and distributed as efficiently as possible became our top priority.
As the business grew 40% YoY from 2016 to 2018 the team was faced with significant growing pains, especially on the supply side, causing increased pressure to assess and rethink some of our core user journey’s and workflows
- Quality reviews: Every artist, artwork that is submitted for sale, goes through a review process, that determines if it is eligible for being listed on the platform or not, which at the time was mostly manual, time intensive and required several members on the business side.
- Fragmented seller journey’s: As the platform grew, each touchpoint i.e. artwork page, sale page, had its own seller journey’s which lead to multiple channels for artwork submissions that needed to be monitored and managed leading to overflowing inboxes and confusion on both the client and business side.
- Endless email threads: Selling artworks is not as easy as selling something on Ebay or Etsy and having the right information is key to accelerate the selling process, particularly around certificates of authenticity, artwork provenance, any information around dimensions, framing etc. Since most submissions started with basic information around just the year, title, artist name etc, members on the cataloguing side of the business team had to in certain instances follow up with countless emails, phone calls and Dropbox links to gather all the information needed to create a listing, leading to significant frustration for the client as well as cataloguer.
Audience
Artnet Auctions serves a wide range of users, from collectors who are just about to start their collecting journey, to dealers and art advisors who professionally buy and sell art, and in most cases, act as representatives of their clients.
One of the great design challenges for this product is that it needs to work for users with a wide range of level of sophistication and expertise, while also taking into account important internal workflows such as the review and cataloguing processes which still require human supervision and checkpoints.
Our initial hypothesis for this project was that if we unify the existing selling journey’s and capture submission data in a more structured way, we would streamline operational workflows and grow the supply side of the marketplace, due to faster review and listing speeds.
My role
I led the project as a product designer on the team, as well as led the UX research in close collaboration with our data science team. I also collaborated closely with auctions Product Manager, lead engineer and content strategist.
Design process
To refine and validate our initial hypothesis we kicked-off the project with an initial research phase, which consisted of qualitative and quantitative research sessions in the form of structured interviews with target users and key members of the business team.
As a first step we sent targeted surveys and analyzed the various funnels in GA in which we looked at current flows through the lens of the seller, analyzing drop-off points, event tracking data, as well as reviewed reports from our internal CRM and inventory databases to get complete picture.
From there we defined our first set of key metrics, which were the overall amount of artwork submissions that would make it to a specialist and their acceptance rate. These would be instrumental to test and validate our initial hypothesis we defined earlier on.
I started the competitive analysis with a feature analysis with annotated screenshots that referenced direct competitors experiences and feature sets regarding the seller journey in the art market. Some of these competitors included, auction houses and some newly entered players in the digital space as well as indirect competitors in the fashion resale space, which closely resembled the workflows of our own audience in many areas as well as eBay’s Valet service for high-end watches.
Throughout the research I worked in close collaboration with the lead PM and data scientist,
leading most of the sessions and making sure we gathered enough data points and feedback in order to start the design process. It was imperative that we had enough information gathered to feel confident about our decision making and hypothesis validation process down the road.
It was also important to be as transparent and collaborative in our work as possible, especially in the way we shared our findings and communicated progress with our counterparts on the business side. Weekly status syncs and a dedicated Slack channel were created with links to all the documentation and this was vital to the success of the project.
After a few weeks we were able to paint a detailed picture of the journey, defining the problem and outlining the key challenges, which led us to define three key goals that we would hold each other accountable throughout the project:
- Think of pages as workflows rather than features, to consolidate selling touchpoints across the site into one unified experience
- Make the selling process simple but not simpler, so that beginners as well as professionals will be able to sell seamlessly and effortlessly, especially on mobile
- Enable a continuous selling cycle of an artwork on the platform, by creating seamless integration with our various systems and workflows
At this point in the process I kicked-off the design phase, starting out with mapping the core user journey’s, diagramming key user flows and sketching out initial versions that I then turned into initial concepts.
Design critiques and check-ins with our team and key stakeholders enabled constructive feedback loops and guided my design thinking which led to concepts I felt would be strong enough to be presented to users for testing and validation. We tested the concepts via Invision prototypes and in later iterations in HTML/CSS with live data. During this phase I worked closely with our engineering lead and data scientist to be in complete alignment about any technical constraints and data requirements for the project and acted as a mediator alongside the lead PM to make sure we were still heading into the right direction.
Design options, decision making
There were two distinct experiences that I created based on the research findings and feedback collected throughout our research.
The first concept was designed as a single page to be completed in one session with a strong focus on vertical flow, without making the user feel too overwhelmed. Forms and other input elements were carefully structured to guide the user through the journey while making sure any of the more advanced fields, which needed further explanation, were accompanied by short help text that would add additional clarity to more complex fields such authenticity questions and measurements.
The second concept direction divided the experience into several distinct sections, combining the various input fields and selectors into groups separated out into multiple pages. This direction’s goal was to reduce the users' cognitive load and make the experience less overwhelming. A stepper UI element would signal to the user where they are in the submission process and when a specific section would be completed, with the intent to invoke quicker rewards and a feeling of success.
Both experiences were designed with mobile as a main use case, since specialists or the marketing team would send out invitations to consign with artnet auctions mostly via email or personal outreach which were opened predominantly on mobile.
Throughout the design process I would create quick Invision prototypes and share them with users from our auctions research group, to gather feedback and further iterate. Our research and strong feedback resulted in aligning on the multi-page experience.
After sharing our findings and recommendations with all stakeholders and key decision makers, we moved forward to build the second concept as the MVP, with the goal to gather real data in a limited roll out to select clients. We would then continue to make adjustments (as needed) within two sprint cycles after it was first released, eventually releasing to everyone on the platform.
Constraints, limitations that had to be considered
Although artnet Auctions launched in 2008, it still operated on an older tech stack than most younger competitors since it was integrated into the overall artnet platform and codebase. This lead to limitations in respect to the speed at which developers could build upon it, as well as introduce risk around breaking other areas of the site by any potential code changes.
Additionally we had to be careful to not downplay the importance of our in-house experts as part of the overall authenticating and value appraising process of incoming artworks. Balancing automation vs human supervision was something that we’re still calibrating today and which will not go away despite advancements in machine learning and data management.
A further limitation that needed to be considered was that of our account section which wasn’t updated since the platform launched and saw only minor improvements and fixes.
Design validation, metrics
The introduction of the new seller experience had a very positive impact on the overall business and we were able to validate our initial hypothesis. By introducing basic acceptance criteria as mandatory fields at the beginning of the consignment flow, the supply side of the marketplace grew significantly faster due to the new time time to review and list. Within the initial launch phase we were able to drastically reduce the amount of steps the business team had to take to review and list works, as well as were able to grow our list of inventory for sale significantly.
Users who experienced lengthy and most often frustrating processes were now able to submit their artwork within a few minutes and receive a response for a potential sale within a few days. We were able to significantly increase the overall value and volume of incoming artworks for sale, helping the team to maintain and increase their already aggressive goals for their upcoming quarters and accelerate the overall growth trajectory.
The most important takeaways from this project were that user experiences should reflect the real world ‘jobs-to-be done.’ They should be the invisible hand that guides efficient workflows, in this case selling artwork quickly and with confidence.
To comply with my non-disclosure agreement, I have omitted and obfuscated confidential information in this case study. All information in this case study is my own and does not necessarily reflect the views of artnet.
