The problem

25% of users abandoned checkout with faced with force account-creation - costing ~$30M/month.

I was asked to reduce drop-off by letting guests complete purchases.

The Shift: Solving the right problem

We launched a direct to checkout flow to quickly meet the brief. However, I paused and asked: was I solving the right problem?

With further research, it revealed that friction wasn’t the real blocker - trust was.

I reframed the problem and led improvements to form UX and strategic flow design.

A/B testing strategy

We tested flow changes across two rounds:

• Round 1: Trust-building form improvements

• Round 2: Registration strategy - button-based login vs streamlined login

Mutivariant test #1: Optimized flow vs direct checkout

Here’s how I exceeded project goals within a month. I broke down the project into two phases

1. Phase one: focused on removing the login barrier and reduce exit rate and run multivariant testing

2. Phase two: focused on optimizing checkout experience.

Exploring viable approaches

After some explorations, I came down to three different approaches.

1. No login screen, takes user directly to guest checkout.
   This approach is the fastest way to checkout, and the ROI will likely be very high. However, this approach lacks prominent support for the existing users.

2. Streamlined interstitial
   This approach offers everyone a way to move forward. Specifically, returning user have a faster and streamlined way to log in by directly enter their credentials. Guests also have a clear CTA to checkout.

3. Button-based interstitial

   Different from the streamlined approach, this design takes on a more basic layout. Simply providing three different CTA for different types of users to click.

Setting up the test

Working with a PM and BI, we set up the testing environment. We released to all anonymous users in 25/25/25/25 split, and ran the test for 2 weeks.

These are the metrics we wanted to measure: Cart exit rate, user registration rate, and order conversion rate.

Multivariant test result and metrics tradeoff

We ran the test for 2 weeks, and here’s the result:
Variant 1: lowest cart exit rate, highest order conversion rate
Variant 2: highest user registration rate

The question we then becomes 'Immediate purchases or long term customer value?’

I set up a discussion with the Business partner, and we’ve decided to move forward with variant 2 because it aligns with Mercari long-term goal to prioritize loyalty and retention.

Checkout optimization

Now that we’ve addressed the barrier and enabled guest users to checkout, doesn’t mean they will at an acceptable rate. Guest checkout users are fickle. We haven’t develop a significant understanding and loyalty. Because of that, any small shortcoming will be magnified.

Here are the areas I optimized on to improve user trust and streamline process.