If you tried everything but just can’t get more bang for your buck, this article will change how you advertise.
Google’s Smart Shopping campaign is a fairly recent campaign type, which combines standard Shopping with display remarketing campaigns, being delivered across the GDN, Gmail and Youtube.
Acclaimed as one of the most performance-oriented campaign types, its promise is to bring the power of automation and machine learning – using automated bidding and ad placement – to promote the products and maximize the return on ad spend (ROAS).
When it comes to campaign set-up and structure, Google recommends to have only one Smart Shopping campaign containing all inventory products, claiming that “the more conversions a campaign has to learn from, the better” (link).
The issue with this approach is: in an inventory with thousands of products with a high price variation (for example, from less than $10 to over $200) the ROAS range will also vary. It may pull the overall ROAS down, inhibiting the full potential performance as only one ad group can be created per Smart Shopping campaign.
In an attempt to improve one of our client’s Shopping performance, we split the stand-alone Smart Shopping campaign into 3 (2 Smart and 1 Standard). At the end of the third month, we managed to scale the campaigns spend and double the ROAS.
In order to achieve these results, we followed the steps:
1 – Feed Optimization
This is a crucial step, as Google needs as much information as possible from each product to be able to deliver it to the right audience at the right time. Make sure all required attributes are correctly filled and insert all additional attributes that suit the specific product. Don’t forget to use the Custom Labels to input extra information (such as seasonality, margin, top products, etc).
2 – Launch one Smart Shopping campaign
If there’s no historical data, launch one Smart Shopping campaign. After one month, if the conversion volume is high, you’ll get enough data to understand the channel behaviour.
If there’s already historical data from a previous Smart Shopping campaign, you can skip this step and go straight to the analysis.
3 – Analyze the data
Analyze the Shopping performance data. One of the best ways I found was creating a custom report and adding the main feed attributes along with the performance data (impressions, clicks, cost, conversions and conversion value). During the analysis try to identify patterns in the products’ performance (such as similar conversion volume or, even better, ROAS), which can appear by brand, product category, price range, etc.
You may find 2, 3 or even 4 big groups, but make sure they’re big enough. Google reccomends at least 100 conversions in 30 days in order to start a Smart Shopping campaign in a good way.
Create custom labels to identify these products if what you chose isn’t an already existing attribute in the product feed.
4 – Create new Smart Shopping campaigns
According to the segmentation you’ve specified previously, create new Smart Shopping campaigns making sure you are only targeting the segment you want in that particular campaign. When the same products are in two or more Smart Shopping campaigns, Google will prioritize one of the campaigns and won’t deliver the other.
Throughout the month, manage the budget across the campaigns to get the most out of them. Don’t be afraid to exclude products, if necessary.
5 – Extra tip 😉
You can create a standard campaign for the high-spend, low-performance products/categories. Set a low budget and a target ROAS. The campaign will be triggered when Google’s signals when there’s a high purchase intent for these products.
=====================
Want to learn more about Google Shopping, paid advertising, and digital marketing? Have a look at the below resources.
• Is your business profitable? Using ROI and ROAS to measure ad performance
• Does Facebook work for your business? Learn how to “read” Facebook data
• Want to run google shopping ads to scale your business? Learn how to create your own product feed
• Check out our latest blog posts
