Case study – Comparative analysis: Third-party CSS VS Producthero CSS in combination with Products AI

Challenge: Enhancing product titles and precision in categorisation

Very Ireland is one of the country’s leading online retailers, operating under The Very Group. With a diverse product range spanning Fashion, Electricals, Home, and Sportswear, they continuously update their inventory to meet customer demand and drive new business. However, optimising product titles at scale is a significant challenge—especially when refining Google’s product categorisation at a highly detailed level.

To tackle this, Very Ireland partnered with Wolfgang Digital, whose team recognised the potential of Products AI—one of Producthero’s latest AI-powered tools—to streamline product title optimisation and categorisation for greater precision.

In this process, Google’s product taxonomy plays a vital role. This structured system classifies products into hierarchical categories, directly impacting ad performance. Accurate categorisation enhances ad targeting and improves the user experience, making products more relevant to search queries.

However, the real game-changer in Shopping Ads is the product title. It influences search visibility, relevance to queries, and even auction competitiveness. This made the challenge twofold:

1. How can product titles be further enhanced while refining Google’s categorisation to the most granular level?
2. How to accurately measure the true impact of Producthero’s AI on performance

With these questions in mind, Wolfgang Digital team sought solutions that would push product optimisation to the next level.

Approach & implementation

“At Wolfgang Digital, testing is in our DNA. Every decision is backed by data, so we structured a controlled experiment to isolate the impact of the Producthero AI tool.” 

Test Setup

To accurately measure the real impact of Products AI, Wolfgang Digital’s team set up a fair and controlled experiment. They launched two identical Google Shopping campaigns—featuring the same products, setup, and start date:

• Campaign 1: Used existing product titles and Google’s product categorisation, managed through a third-party CSS partner.
• Campaign 2: Used AI-generated product titles via the Producthero CSS, rewriting 1,471 titles while refining Google’s product classification at a more granular level.

To ensure a fair comparison, they kept the setup simple: third-party CSS vs. Producthero CSS. This approach eliminated any CPC discrepancies, making it a true test of AI-driven improvements.

They also chose to run the experiment on Shopping campaigns rather than Performance Max. Why? Because Shopping campaigns allocate 100% of the budget to Shopping ads, whereas Performance Max distributes spend across multiple channels. This kept the data clean and reliable.

The experiment ran for four months, focusing on the Home category. To push Producthero’s AI to its limits, they used the “Hard Goods” template with AI Creativity set to High—challenging the AI to rewrite and optimise product titles for maximum impact.

Results

• +11% CTR increase
• +5% ROAS increase
• 1,471 products were optimised.

Although both campaigns launched simultaneously with identical setups, there was one crucial difference: Google Merchant Center retains historical product data. One campaign ran through a third-party CSS, benefiting from existing data in the Merchant Center, while the other used the Producthero CSS, starting from scratch with no historical advantage.

In theory, the third-party CSS should have had a clear advantage. But the AI had other plans.

“What’s even more impressive is that the product titles were already well-optimised before the test started. Yet, AI still managed to boost performance even further.” — Otavio Dias, PPC & Analytics Specialist, Wolfgang Digital.

As the CTR increased, ad relevance followed suit. By the fourth month, when the influence of historical data likely became less significant, the campaign using Producthero CSS experienced an 11% boost in CTR. This resulted in an 18% increase in ad spend.

Despite the higher spend, ROAS improved by 5%, showing that the increased relevance of the ads had a positive impact on the overall performance. A higher CTR signals to Google that the ads are more relevant, which can ultimately lead to greater business growth.

 

Future actions

Given the significant impact of Producthero’s AI tool, Wolfgang Digital team is now focusing on scaling the optimisation across an even larger range of products in Very’s extensive catalogue. The goal is to maximise both efficiency and performance across all categories, ensuring continued growth and success.

Client testimonial by Wolfgang Digital

Producthero’s AI took already well-optimised titles and made them even stronger – even when we were at a disadvantage for a short period due to the third-party CSS from the A/B test containing historical data advantage. That’s seriously impressive. We’re excited to keep refining our shopping ads as they’re the backbone of any e-commerce growth strategy.

About Very.ie 

• Very is Ireland’s leading online retailer helping customers to get more out of everyday. With up to 70,000 products online across 1,800 brands, Very is the online store for fashion, home & more.
• Originally launched as Littlewoods Ireland in 1923, rebranded to Very Ireland in 2022
• Headquarters in Dublin (Ireland)
• https://www.very.ie

About Wolfgang

• Wolfgang Digital is Europe’s most awarded digital marketing agency, providing best-in-class Paid Search and Social, SEO-Content and Email Marketing
• Founded in 2007
• Headquarters in Dublin (Ireland)
• https://www.wolfgangdigital.com/ 

About Producthero

• Software vendor for Shopping Ad tools
• Comparison Shopping Service (CSS Google Partner)
• Headquarters in Amsterdam (The Netherlands)
• Producthero.com