Manually Selecting which Products to Display

Manually Selecting which Products to Display

Thomas Byskov Madsen & Lasse Rasmussen - November 3, 2017

An interesting option that comes with the SPAG Structure is that it enables you to manually select which products to show for specific search terms. However, whether this should be done or not is a frequently debated question. To shed light on it, let’s return to the red t-shirt example and assume that we sell anywhere between 1-10 different red t-shirts, depending on our stock. Now assume that three of these red t-shirts sell significantly better than the remaining seven. Should we then negate “red t-shirt” on the seven low-performing products, thereby ensuring that we will only show our top sellers for this search term? One obvious reason for doing this is the expected increase in conversion rate gained by only showing the most popular products. However, there are also several compelling arguments for why we should not be doing this. These include:

 

Manually selecting products to display

These risks might seem manageable on a small scale with 10 red t-shirts. But assume that you are running Google Shopping on a webshop with 10,000 or even 100,000 products. Keeping track of which search terms you have negated for which products can then become overwhelming.

Furthermore, as a Shopping campaign matures and accumulates a significant amount of transaction data on individual products, top-selling products will start to stand out naturally. Through bid management, maximum click prices on these products is likely to increase, and this, combined with a CTR-level that will probably be higher than on other products due to popularity differences, will ensure that top-selling products are displayed most often.

Given the risks above, and the fact that bid management provides an effective solution in the long run, we at QuantAds do not recommend following this strategy – unless you have a very small, easily manageable product selection. In general, negative keywords should only be used for traffic segmentation and excluding search terms that are generally irrelevant or consistently underperforming.