Vinted.com have a bounce-rate issue on their home page and they want to decrease it.
Note: I did not work for Vinted, all the details below are purely conceptual.
Vinted has a 64.79% bounce rate on its web home page, more specifically for people coming from its FB campaign.
So, vinted.com rushed to me, almost crying, begging me to solve their pretty high bounce rate. I decided to help them.
First things first,
Is it a high bounce rate? Well, 45-60% is considered average, but here’s the thing, their first-time visitors were coming from an FB ad with a high intent to buy.
After a quick UX audit, we identified a bunch of potential bets we can place.
Started with this market engagement hypothesis:
If we manage to show more relevant items to our first-time buyers, the activation rate will skyrocket.
Then made it a bit more actionable like:
At least 20% of new users will click on an item that is for sale if they see items they are interested in.
Finally, transforming into an actionable bet:
At least 20% of new users coming from an FB ad showcasing some trending items will make an offer or add it to a wishlist if we only show them more items they are actually interested in.
Small note: We choose the one above based on its potential impact and effort.
Now let’s prototype it:
When a user lands on the home page, Vinted nudges them to define their interests in order to customize their home feed. And you need to strike the balance of how long you’re going to ask them questions before you try and deliver value. Because it is a funnel, right? And the more questions you ask the more there will be drop off.
How do you actually test it?
Well, you do not have to actually develop it, just make a prototype and do an A/B test. Build an HTML/CSS version of it with no functionality and put it out there for some % of the traffic of first-time visits.
Action plan:
Prototype the personalized onboarding
Put in the browser (write some HTML/CSS for it, no back-end)
A/B tests it with 20% of the first-time visits.
Results:
In a real-world scenario, this data would be gathered from web analytics tools like Google Analytics, customer relationship management (CRM) software, or other data analysis tools.
However, I can describe a hypothetical scenario to illustrate how one might estimate the improvement:
Before Personalized Onboarding:
Total new user sign-ups in a month: 1,000
Number of users who made a purchase within the first 30 days: 150
Pre-Personalization Activation Rate: 15%
Analyzing the Improvement:
Improvement Calculation: The activation rate increased from 15% to 25% after implementing personalized onboarding.
Conclusion:
In this hypothetical scenario, the personalized onboarding had a significant positive impact, increasing the activation rate by 66.67%. This suggests that new users found value in the personalization and were more likely to make a purchase, indicating an improvement in user engagement and a potential boost in long-term retention.
Ok but what about if this is not working? Did we pay for nothing? Not really, here are some valuable takeaways from an experiment where the activation rate did not improve after analyzing user behavior data:
Now we have a better understanding of user and market expectations, behavioral insights, risk mitigation, and so on…
Each experiment provides a learning opportunity and, even when the results aren’t as expected, they contribute to the overall understanding of the users and the product. This iterative process is fundamental to product development and continuous improvement, leading to better user experiences and business outcomes over time.
Leave a Reply