---
title: "Politikos"
url: "https://www.omnicliq.com/en/case-studies/politikos-ai-product-scoring/"
description: "How AI scoring of every product from GA4 signals sent Politikos's ads to what was selling, during the January sales."
category: "ecommerce"
client: "Politikos"
datePublished: "2026-10-28"
author: "Dimitris Bachtsevanis"
locale: "en"
---

# Politikos

> How AI scoring of every product from GA4 signals sent Politikos's ads to what was selling, during the January sales.

## Project

**Client:** Politikos
**Industry:** Fashion
**Duration:** January 2026
**Services:** ML & AI, Google Ads, Analytics
## Key results

- **+49%** Revenue
- **+39%** Ad revenue
- **+58%** Orders
- **+83%** Search value
## More sales in the January sales, with no time for slow decisions

Politikos is one of the largest Greek fashion e-shops, with thousands of products from well-known brands. In the January sales, demand shifts day by day by category, size, stock and price.
The goal was more revenue and more purchases than in January 2025. The hard part was not increasing ad spend but deciding which products deserve it.
In the sales, last month's data does not help: a product that sold well at full price may be left without sizes, while another suddenly becomes a bargain once discounted.
With such a large range, manual analysis was too slow. By the time it was clear which products sell, much of the period would be over and budget would have gone to products with no momentum.

## Every product gets a score from what visitors do, every day

We built a system that reads from GA4 what visitors do with each product: views, add-to-carts, purchases, revenue and category trend.
A forecasting model (MindsDB) turns these signals into a commercial momentum score for each product. It follows no fixed rules; it spots an upward trend before it shows in sales.
The score does not stay in a report. It flows automatically into the product feed as a label, so Shopping, Performance Max and Search prioritise the products that are rising.
The score refreshes daily, so a product that runs out of sizes loses weight immediately and budget moves elsewhere. Time from data to decision dropped from weeks to days.

## Results

- **Revenue** — More revenue in the January sales, without materially more visits.
- **Google Ads revenue** — More revenue from Google campaigns, which went to the highest-scored products.
- **Orders** — More purchases from demand that was already there.
- **Search value** — Sales value from Search campaigns almost doubled.