Brand searches are the most valuable traffic source for every brand. In Google ads, the highest sales volume at the lowest cost also comes from brand searches. But once the volume of this traffic source reaches a certain level, a heavy media investment is needed to increase it further. So if we couldn’t make that investment, how could we increase the revenue return of advertising?
Our goal, in a period of intensifying competition on Google Search, was to lower the cost of sales and double online sales revenue without increasing media investment. The question on our minds was: how could we shift the dominant share of brand searches in revenue to other channels?
Strategy & Execution:
We started by examining our decision tree algorithm, which processes Google Analytics and CRM data. According to our analyses, “Google Shopping Ads” were performing well below their potential.
Put simply, in this ad model, where products are sent to Google via an XML data feed, we used A/B tests built for an e-commerce site with about 7,000 products to prepare a database containing more than 100,000 variations. We knew that users’ search behavior was shifting every day from one- and two-word searches toward searches containing multi-word phrases. Using this insight, we multiplied our product listings with keywords including size, color and pattern. Ads appeared for users specifically for products matching the size and color they searched for, and we avoided unnecessary ad costs for colors and sizes that were out of stock. The convenience and experience we offered users also raised our quality score.
The area where we determined the best-performing product listing feed by testing Google’s algorithm was the product price information field. The data feed that sends product information to Google contains 3 different variables: “list price,” “sale price” and “price.” The promotional labels in Shopping ads were also automated depending on how these variables were used in different ways.
We varied these three variables in different combinations for 7,000 products. Based on this work, we tested user response to the varying “sale price” and “price drop” ad labels.
Results:
In an environment where many brands emphasized “discount” and “old price” in their campaign announcements, we discovered that the “price drop” label performed much better. With the right variations we reached through data analysis and testing, we both gave the most accurate response to users’ search behavior and set ourselves apart from our competitors.
- We reduced the cost per click of the ads by 62%.
- We increased the product sales rate by 187% and return on investment by 211%.
- While raising the share of Shopping ads in total revenue from 5% to 67%, we broke the dominance of brand searches and made Shopping the highest-revenue channel.
- At the end of the period, we exceeded our target revenue by 50%, achieving a 200% increase over the previous period.

