>> Variant B improved conversion rate (CVR) or Add-to-Cart (ATC) rate? Roll it out.
Basing rollouts on add-to-cart rate is generally not a great idea, particularly for experiments on product pages. There are a variety of reasons why shoppers add products to carts that may only be loosely based on intent to purchase. In my years of ecommerce A/B testing, I've seen plenty of tests that improve conversion rate and/or revenue per visitor but negatively impact add-to-cart rates.
doctorpangloss 2 hours ago [-]
> that hit me like a ton of bricks
> _mechanics_—how to change a digit on a screen—while
> isn't just a failed test; it’s a solvency risk.
if I wanted to ask ChatGPT how to do something, I would
augmachina 3 hours ago [-]
"Local area person discovers economics.Confidently Explains Everyone's Problems Now Solved"
mrkaluzny 3 hours ago [-]
What would you prefer to see on this topic?
qingcharles 3 hours ago [-]
I thought it was an excellent article myself. Very thorough.
mrkaluzny 3 hours ago [-]
Thanks! It's hard writing into void any feedback is highly appreciated ;)
xiphias2 3 hours ago [-]
For doubling success I suggest ruining the product slowly and testing how crappy the product can still be while the customer is still paying for the brand (as he's expecting some improvement for the higher price), and voila, we've got enshittification.
Rendered at 19:18:35 GMT+0000 (Coordinated Universal Time) with Vercel.
Basing rollouts on add-to-cart rate is generally not a great idea, particularly for experiments on product pages. There are a variety of reasons why shoppers add products to carts that may only be loosely based on intent to purchase. In my years of ecommerce A/B testing, I've seen plenty of tests that improve conversion rate and/or revenue per visitor but negatively impact add-to-cart rates.
> _mechanics_—how to change a digit on a screen—while
> isn't just a failed test; it’s a solvency risk.
if I wanted to ask ChatGPT how to do something, I would