April 23, 2021
Three Approaches to Shopping Segmentation in Smart Shopping
Google Shopping ads have become an increasingly important portion of the Digital Marketing Mix over the past few years. As such, Google is continuously updating their products associated with this channel. Advertisers who neglect this channel are missing out on a major opportunity to reach their customers with a dynamic, highly-clickable placement that appears above the fold for all relevant searches.
This is the second of two posts on The New Shopping Best Practices – focusing on Shopping Attributes and Feed Adjustments. You can find best practices for Shopping Attributes and Feed Adjustments here.
Google Smart Shopping is an inarguably powerful product. It relies on large amounts of data to drive performance. It’s here to stay. It is not an excuse to dump all products into one generic “Smart Shopping” campaign. Moving away from a Shopping Structure where all products are grouped into one Smart (or worse, one Standard) campaign can yield huge gains – improved performance is almost guaranteed if you upgrade from no structure to a goal-driven segmentation that aligns with larger business strategy.
There are several approaches to Shopping Structure for both Smart and Standard Shopping formats. Because this is extremely KPI driven, no single approach qualifies as a best practice for all situations.
Keeping that in mind, a universal primary consideration when planning a structure is the balance between Data Aggregation and Spend / Return Control. This is an important consideration for both Smart and Standard Shopping accounts:
Search Term Filtering in Shopping uses negative keywords to sculpt search traffic. It allows us to breakout out user groups into 2-3 separate budget / aggression categories:
Each level of the filtering structure carries full product coverage, using a combination of campaign priorities and negative keywords to split traffic between the three levels. Remember,
shopping campaigns use Product Titles for keywords, so we cannot control this at a campaign keyword level.
This account is breaking out larger Product Types at the Ad Group level, which makes for a clear example:
Top Funnel searches that our products are eligible for are covered by the Generic campaign. Specific High-Intent NB searches land in the Part Sizes campaign, and Brand searches land in the campaign with that name. If a client’s business is not conducive to High-Intent NB searches, this can also be run with Top Funnel and Brand Filtering alone.
This is the most effective way to allocate spend between separate larger user groups in a Standard Shopping structure. If interested, a deeper dive on Search Term Filtering can be found here.
Despite the performance improvements found by filtering Brand searches into a separate campaign, Google strongly recommends against filtering out any searches from Smart Shopping efforts. Why? The logic should be the same between the two formats, right?
Instead of blindly assuming this strategy wouldn’t work in Smart Shopping – specifically when it has been a strong performer in Standard – we tested it to find out. The results were clear:
This test was run in a high-volume campaign, outside of Holiday or COVID seasonality. To preserve confidentiality only % lifts are provided.
What we learned from this test is that Google Smart Shopping needs access to Brand searches to drive its machine learning algorithm. This system learns which user types are highly valuable / likely to purchase from Brand searches (which are by nature, high-intent), and then monitors for similar user signals across Non-Brand searches. This enables it to continue driving high performance at increasing scale.
By removing Brand searches from the Smart Shopping algorithm, we crippled its ability to learn from a highly important user group.
Google’s Smart Shopping is a powerful product – it automatically optimizes budget towards the highest value user groups by tracking user signals and their likelihood to convert, within a machine-learning algorithm. This system is highly effective, and has driven demonstratable performance improvements across a variety of clients. It also occasionally makes mistakes.
Here is a rough timeline of what happens when Smart Shopping processes a search:
The area where Smart often misses is deciding not to show an impression. While the systems have the ability to process huge amounts of information at a scale humans cannot, they are vulnerable to missing out on potential customers due to their lack of ability to “think” outside of the historically gathered information they use to process bids.
To ensure we always show an impression whenever eligible, advertisers should consider running Standard Shopping campaigns behind their higher-priority Smart Shopping efforts:
In the example above, Standard Fallback Campaigns are found in the box. These fallbacks are using a simplified Search Term Filtering structure that separates out Non-Brand and Brand searches. At the low volume we typically see for these campaigns, there isn’t enough traffic to optimize for a High Purchase Intent group of campaigns.
Due to restricted volume, this tactic is extremely low-touch compared to most other efforts.
In the example structure above, the Standard Shopping Fallback campaigns picked up an additional $140,000 in sales In lower volume accounts than this one. In general, we expect to see 5-10% lifts in annual revenue from these fallbacks.
Google Shopping ads have emerged as an inarguably important component of an ecommerce marketing mix – this format can truly elevate your paid search efforts. Many retail advertisers have begun considering this as its own separate channel, allocating specific budgets to Shopping (and Feeds Management), due to the additional revenue streaming in from Shopping.
If you are a business looking for help with your shopping campaign strategy, please contact us.