+27% revenue YoY.
How an AI model scored every product in G for Gadget's huge catalogue, so ad spend went to the products that bring revenue.
More revenue from a catalogue nobody can read in time
G for Gadget sells technology and gadgets online across many categories, with tens of thousands of products. The goal was more revenue from product ads.
No team can judge by hand which product deserves exposure in such a range. Many products get few visits, so there is little data to judge them individually.
Without priorities, ad spend is spread evenly and lost on products that do not sell.
Platform automation usually focuses on a few well-known best sellers and leaves most of the catalogue unseen, even when it has demand.
A model predicts which product will sell
We connected each product's GA4 behaviour (views, add-to-cart, purchases, demand trend) with feed data (category, brand, price range, availability).
A machine learning model (MindsDB) gives every product a score for its likelihood to drive revenue. For products with little data, it learns from similar products in the same category and brand.
The score becomes a feed label, and Google Ads product campaigns prioritise the products most likely to sell.
Scoring runs daily across the whole catalogue, so a product that starts attracting demand gets exposure immediately.
From behaviour to score
Signals per product
We gathered visitor behaviour for every product and joined it with the commercial attributes in the feed.
A score from the model
The model learns from sales and scores every product, so priorities come from data rather than gut feeling.
Priority in the campaigns
The score reaches the feed as a label and the Google Ads campaigns push the products most likely to sell.
The Results
Revenue
More revenue than in the same quarter a year earlier.
Average order value (AOV)
Each order is worth more, because the ads go to higher-value products.
Orders
More purchases, from visitors with stronger buying intent.
Products scored
The number of products the model scores every day, across the whole catalogue.
Too many products to know which ones sell?
We score every product with AI and send your ad spend where it brings revenue.
Start a conversation