Buyer acceptance rate at OLX

Year

2022

My role

Design

Team

PM, Data Analyst, Engineers

What's been done

Moved with small iterations, designed a minimal solution that was shipped in 2 weeks, then created a more substantial solution for better UX.

Context

When buying on OLX you can either pay online right away or pay when getting the item at the post office. Sellers have less trust in purchases made with after payment since buyers "aren't invested yet" and "will probably reject the purchase".

Checkout screen — We've started with phase 0: blocked the abusers of OLX Delivery

There was a small number of buyers that were abusing the system by either ordering multiple items only meaning to pay for one that was best. This was the fastest/easiest solution to block them and start thinking about how can we make this system better.

Acceptance rate states — Our idea was to give a sort of rating for buyers that usually visit the post office and take their order versus ones that mostly ignore their purchases or usually reject them.

Deal confirmation screen — Here's how that would look for the seller. When confirming a deal they'd see how often deals end positively for this buyer.

This would give more trust to sellers since they’d see that even though this is an order without a online pre-payment, the buyer has good reputation and usually finishes the deal.

Acceptance rate screen — Buyer can open a page in their profile to see how they're doing. They also see a preview of how they look for the seller.

I’m very mindful of all kinds of error states in apps since it isn’t always obvious that you might have problem with the connection.

Acceptance rate desktop screen — Here's how the same screen looks on desktop. Looking back, I'd design it differently. I don't like that there's so much white space.

Different acceptance rates — Now when purchasing buyers see their so called acceptance rate.

Now this is the second iteration of what was done in “phase 0”. If acceptance rate is below the threshold, payment at the post office becomes disabled and immediately explained why and what can be done.

A/B test screens — We also didn't want to cannibalize anything for new users since they obviously won't have an acceptance rate right away.

We did an A/B test where besides the control (nothing) we had a variation where we said that acceptance rate wasn’t formed yet. Control performed better so that’s what we finalized.