<BACK

Seller flow optimisation


Situation

As a Product Manager for experiment team, I had a task to find ways of improving conversion to lister (from a new member to uploading an item, since Vinted is social marketplace for pre-loved fashion).  By joining forces with performance marketing team (responsible for paid traffic) we were able to increase the conversion by 70%.

 

Big picture (mapping the flow)

Identifying steps for new members to list their first item, was eyes opening exercise.  This helped to understand the whole user journey and discover unnecessary friction points. 

 

Metrics (constructing the funnel)

By adding conversion rates to every step, we were able to confirm which friction points caused the biggest drop-off, as well as it sparked new ideas. Funnel analysis also helped to draw the line in the sand and set the goal for seller flow optimisation.

 

Assumptions and Hypothesis

After we had the funnel in place, we brainstormed different assumptions, why members may be dropping off at specific places, what problems they might be encountering that cause them to leave. We did it internally in our team and asked the whole company to contribute their ideas. By looking into flow and numbers, we prioritised ideas by potential impact.

Some of the assumptions and hypothesis we came up with:

  1. As a new member I cannot upload the item, because I'm not at home. If we would remind her when she is at home, she will be more likely to upload.
  2. As a new member I want to see how product and upload form looks before deciding to upload. If we would ask new members to register on the very last step, members will be more likely to decide listing.
  3. As a member I get confused when camera mode appears after I press upload new item. If we would show upload form first, new members will easier understand the flow.

 

Experiments

We selected the most promising hypothesis and constructed the experiments for each of them. Most experiments were as A/B tests. We also performed interviews with members who dropped-off without uploading. It helped to test some of our hypothesis as well as sparked new ideas to be tested.  For each experiment we set a success criteria and potential next steps if experiment would fail. We used Trello to track the progress. 

 

Results

From around 10 experiments we ran, 2 were successful and fully implemented later on. Besides reaching our goal for new lister conversion, we learned about how different parts in the product contribute in acquiring new sellers.