Catawiki
A CASE STUDY
Refining seller registration
A combination of motivation, ability and prompts to improve the registration journey at Catawiki.
PROJECT DURATION
YEAR
6 months
2023
ABOUT CATAWIKI
Catawiki is an online auction marketplace for special objects and collector’s items ranging from classic cars to ancient coins. Being a curated C2C marketplace, these are the three main actors in the journey connecting supply to demand.
CONTEXT
In 2023, Catawiki saw an opportunity to retain about two third of sellers who were dropping off before getting their object approved and placed into auction by an expert. For a secure sale, sellers are required to verify their identity at a few moments during registration. Only a third of sellers were completing verification, making them eligible to sell on Catawiki.
THE CHALLENGE
Enable sellers to complete identity verification so their objects can be placed in auction and sold.
Our high level goals were:
- Make verification feel seamless, less tedious
- Maintain trust, so sellers and buyers feel safe making transactions with us
- Keep verification legally compliant while gathering as little information as possible
MY ROLE
I led the design for the verification experience, collaborating with two other designers on fine tuning the north star.
I worked closely with a researcher, a writer, two product managers and an engineering lead, additionally collaborating with the legal and customer experience team to bring the project to completion.
A CLOSER LOOK AT IDENTITY VERIFICATION
Verification was happening at multiple points in the seller journey. All these steps were required to perform a KYC by Catawiki’s payment partners, Stripe and Payoneer. Here’s a list of the different kinds of information required at various steps before an object could be placed into auction.
We were asking sellers for a lot of information in order to verify their identity before submitting their object to Catawiki. Sellers could choose to skip some of these steps to proceed, but were required to complete them eventually for their object to be placed into auction.
When Catawiki experts reviewed objects from unverified sellers, more often than not, they would send the submission form back for adjustments asking the seller to first get verified. During the review it was also likely for the object to get rejected if it didn’t meet Catawiki’s quality guidelines.
THE DISCOVERY
Verification was perceived as a "long straight corridor"
I leveraged prior user research on this topic to understand the challenges of the current registration flow. The main insight was that seller registration and verification involved a lot of steps, some of which felt daunting for new sellers. Some additional insights were:
- Some of the information required wasn’t signalled upfront, meaning that sellers were caught off guard when they have to provide it.
- Third party authentication felt daunting for new private sellers if they didn’t know what Stripe was.
- Sellers wanted to have clear estimates of how long the process was going to take, particularly those parts of it which are out of their hands, like the expert reviews.
- Sellers new to Catawiki struggled to trust the platform. They didn't always feel comfortable providing their IBAN upfront even though they knew would be used to get paid out.
REFRAMING THE PROBLEM
Expecting heaps of effort from sellers before their object is approved by an expert causes them to give up mid way.
Sellers were required to give us a lot of information before being assured that their object was a good fit for Catawiki. Stripe and Payoneer verification is lengthy and sometimes took a couple of days causing the seller to lose momentum and interest in selling. There were around 12 registration steps a seller had to complete before being able to submit an object and have it approved.
THE REGISTRATION RE-DESIGN
Using a combination of motivation, ability and prompts to enable sellers to provide information bit by bit along the entire journey from registration to shipping.
Simplicity and core motivation changes behaviour
As the Fogg behaviour model suggests, prompts work best when motivation is high and a task is easy to perform. Providing personal information can feel tedious especially when it interferes with the task at hand. Our goal for the north star was to minimise the delay caused by information gathered upfront and instead nudge sellers for it when their motivation to provide it was the highest. Here’s what it looked like.
We started with a simple registration after the seller selects the object they want to sell. Having already selected an object we could showcase other similar sold items to help sellers learn what they could earn. The registration was just a quick sign-up using name and email so sellers can get to listing their object.
Knowing a combination of the seller's country and object, we'd prevent sellers from spending time and effort listing an item which we were not compliant to sell.
The seller's address was requested at a later stage once it was relevant, ie, when the object had been bought and paid for, and it was time for the seller to ship.
IBAN and third-party payment verification was only requested when Catawiki had received the money and it was time to pay the seller. We also helped the seller know what to expect.
At any point in the journey, the seller has the option to complete all verification steps. We placed banners visible to unverified sellers only in between important tasks.
HOW DID WE GET THERE?
Understanding the breadth of personal information gathering
A number of data points help us verify the legitimacy of a seller. These were concentrated in the registration step. It was crucial to understand what the purpose of gathering some of this data was and how it was used by Catawiki or third party verification services. We found that some of the information was unrelated to registration, yet requested from the seller early on. Here’s what the before and after of the information structuring looked like.
Before - A lengthy registration with a lot of information at once
After - Registration blended with object submission and spread across other parts of the seller journey
Defining design principles for verification
Based on product benchmarking and user research I defined a set of principles that guided the design and communication for the north star.
Testing the north star with 6 new sellers
Main insights we took into consideration for improvements:
- Users were skeptical regarding the impact of delayed verification, especially for the buyers.
- Information on the flow and some parts of the email informing sellers of next steps remained unnoticed and unread.
Improving communication along the seller journey
Prompts and communication were to play a huge part in the success of this project and so I took the opportunity to re-define email messaging in collaboration with the UX writing team. I conducted a small workshop to review all the key moments of email communication, re-design some and introduce a few new ones where necessary.
Working around the legal implications of moving payment verification towards the end of the journey
Research showed that sellers were most motivated to complete payment verification when it was time for payout. However, moving a crucial step of verification to the end of the seller journey meant leaving room for sellers with high risk profiles to make a sale without being verified. We worked with the legal team to define high risk cases and direct them to a different flow as compared to the north star. Here’s what the risk analysis and flows looked like.
Exact numbers omitted for confidentiality reasons
Using data to weigh out consequences of shipping before vs after verification
Preventing sellers from shipping before getting verified meant a marginally better seller experience as they would still have ownership of their object in case they get rejected by third party verification services. However it meant a much worse experience for a buyer who has won and paid for an object and needs to wait endlessly for delivery.
On the other hand, allowing sellers to ship meant that we’d be holding their money if they had trouble getting verified. The buyer would not be impacted. I worked with our data scientist and legal team to find out that a very small amount (data intentionally omitted) of hard rejection cases occur every year. These are typically because of fraudulent accounts. Currently this number is calculated before the expert reviews objects. In the north-star, since verification will happen all the way towards the end, the review process would also weed out fraudulent sellers. Based on that learning we chose to let sellers ship before being verified by our third party partners.
ROLL OUT
Working backwards from the north-star
The path up to the north-star was divided into smaller milestones to achieve maximum impact early. We plan to move verification all the way towards the end of the journey in three milestones.
The first milestone is complete and allows an unverified seller's object to get approved creating motivation to get verified. It was rolled out as an A/B test the goal of which was to signal improvement.
The second milestone will allow the seller's object to get sold without needing to go through payment and third party verification.
The third will minimize registration steps and allow more objects into the top of the funnel after one and two would have improved the bottom of the funnel.
IMPACT
Approving an object before verification signaled improvement
8.3% increase in sellers who got their first object approved.
5.4% absolute increase in sellers who's first object went to auction.
These were both compared to the 2023 trend for the metric mentioned. An increase in approvals and objects in auction for first time sellers signalled that registration was less of a hurdle.
REFLECTING BACK
My top takeaway
Time invested early, communicating the vision to rally the team and foster a sense of ownership, pays off later. The engineering team had a solid understanding of the project because of their participation in design feedback sessions early in the project.
One thing I would do differently
I would double the anticipated time needed for legal requirements and start collaboration with the legal team a step sooner.