In-Campaign Traffic Segmentation

In-Campaign Traffic Segmentation

Thomas Byskov Madsen & Lasse Rasmussen - November 3, 2017

Besides segmenting traffic types, it is also possible to segment identical search terms based on the individual products that they trigger. This is done via a method we at QuantAds call the ‘Single Product Ad Group Structure’ or simply SPAGs.

The Spag Structure

As the name indicates, the SPAG Structure means that every single product in the feed will have its own ad group. This is in contrast to having all products placed within one ad group or creating ad groups based on product categories or product popularity, which are some of the typical approaches used in Google Shopping.

Comparison of Google Shopping Campaign Structures


Google shopping campaign structures


The biggest advantage of the SPAG Structure versus other approaches is that it enables us to evaluate search term data for every single product. As search term reports are only available at ad group level, this is not possible in ad group structures where multiple products are placed in the same ad group, as data is then aggregated. In the example below, reports for a particular search term are shown according to the two different setups. With all products in one ad group, each unique search term only appears once, and is aggregated into the single ad group. In the SPAG Structure however, the same search term can be assessed in many different ad groups, all representing a unique product.


SPAG structure


This segmentation enables us to determine how a specific search term performs in relation to the different individual products. With this information, we can choose to exclude the search term (via negative keywords) for one product, but keep the search term triggering ads on other products, depending on performance. The SPAG Structure thus provides us with an indirect financial advantage, as we can limit the amount of products we want to have shown on certain search terms, thereby saving money if they are not performing as desired.

The primary drawback of the SPAG Structure is that it increases the complexity of the Shopping campaigns, making them harder to manage manually. Furthermore, there will always be a part of the product portfolio that represents a very limited dataset. Here, bid management has to be done in a different way, as there won’t be an ‘Everything else’ product group like in most Shopping campaigns, where one bid can be set based on the aggregated data of all the products with limited data. Thus, in the SPAG Structure each product has to be given a specific bid. This challenge can be dealt with in two ways:

1. Setting the same bid for all the product groups with limited data, based on either 1) an aggregation of products with data below a certain threshold, or 2) data from the entire campaign. These bids will need to be updated manually, and as product groups begin to get sufficient data, they should then be given an individual bid based on performance.

2. Using automated bid engine software that as in scenario 1 calculates bids based on either all products with limited data, or all products in total. The bid engine should then update bids individually on each product group, as they begin to accumulate sufficient data.

How to handle bid management for Shopping campaigns will be treated in depth in the third part of this white paper series.