Community Discussion · Policy

Boosted ROI from 0.8 to 3.2 Using AI for Product Selection: Here Are the 3 Steps

Fan MengyaoJul 102026/07/10 84 views

I used to pick products based on gut feeling, and it hurt watching the ad spend burn. Later, I switched careers to learn AI, spent two months tinkering, and now use this toolkit to save me 4 hours of product selection time daily, with a real conversion rate boost of 50%. It’s basically three steps: 1. Use AI to scrape high-frequency words from competitor reviews to spot demand gaps; 2. Have the model run historical order data to predict hit probability; 3. Let AI auto-adjust bids during ad placement. Now my ROI is positive, steady at 3.2. Not showing off—happy to discuss details.

Policy

1 replies

Ctrl + Enter to reply

Siqi Draws PPTJul 11(edited)

[quote="fan_mengyao, post:1, topic:285"]

I used to pick products based on gut feeling and burned through ad budgets painfully. Then I pivoted to learning AI halfway through my career. After tinkering for two months, this toolkit now saves me 4 hours of product selection time daily, with actual conversion rates up by 50%. The core is just three steps: 1. Use AI to scrape high-frequency words from competitor reviews to identify demand gaps; 2. Have the model run historical order data to predict hit product probability; 3. Let AI automatically adjust bids during ad placement. Now ROI is positive, steady at 3.2. Not showing off, feel free to discuss details.

[/quote]

This process is essentially a shift from experience-driven to data-driven product selection framework. But from the perspective of doing strategy at Huawei, the core competitive moat lies in the granularity of data cleaning and the speed of model iteration. Simply running historical order predictions isn't enough; I suggest focusing on long-tail keywords and competitor trend changes.